decoration decoration
Stories

GROKLAW
When you want to know more...
decoration
For layout only
Home
Archives
Site Map
Search
About Groklaw
Awards
Legal Research
Timelines
ApplevSamsung
ApplevSamsung p.2
ArchiveExplorer
Autozone
Bilski
Cases
Cast: Lawyers
Comes v. MS
Contracts/Documents
Courts
DRM
Gordon v MS
GPL
Grokdoc
HTML How To
IPI v RH
IV v. Google
Legal Docs
Lodsys
MS Litigations
MSvB&N
News Picks
Novell v. MS
Novell-MS Deal
ODF/OOXML
OOXML Appeals
OraclevGoogle
Patents
ProjectMonterey
Psystar
Quote Database
Red Hat v SCO
Salus Book
SCEA v Hotz
SCO Appeals
SCO Bankruptcy
SCO Financials
SCO Overview
SCO v IBM
SCO v Novell
SCO:Soup2Nuts
SCOsource
Sean Daly
Software Patents
Switch to Linux
Transcripts
Unix Books
Your contributions keep Groklaw going.
To donate to Groklaw 2.0:

Groklaw Gear

Click here to send an email to the editor of this weblog.


Contact PJ

Click here to email PJ. You won't find me on Facebook Donate Paypal


User Functions

Username:

Password:

Don't have an account yet? Sign up as a New User

No Legal Advice

The information on Groklaw is not intended to constitute legal advice. While Mark is a lawyer and he has asked other lawyers and law students to contribute articles, all of these articles are offered to help educate, not to provide specific legal advice. They are not your lawyers.

Here's Groklaw's comments policy.


What's New

STORIES
No new stories

COMMENTS last 48 hrs
No new comments


Sponsors

Hosting:
hosted by ibiblio

On servers donated to ibiblio by AMD.

Webmaster
Mike Anderer, "MIT" Scientists, Spectral Analysis, and a Patent Application
Friday, September 30 2005 @ 03:50 AM EDT

Stats_for_all has found something truly fascinating. There is a published patent application, #20050216898, filed September 13, 2004 and just published September 29, 2005, for a "System for software code comparison." One of the inventors is a Michael Anderer of Salt Lake City, Utah. That wouldn't be Darl McBride's old pal, Mike Anderer, now, would it?

You remember him, don't you? Remember his leaked memo with all the misspelled words back in March of 2004 that revealed that BayStar was a Microsoft referral and that Microsoft sent $86 million SCO's way, "including BayStar", thanks to Anderer?

And then a few days later, he wrote a letter which he published on NewsForge about Microsoft's plan to destroy Linux, and he described himself like this:

I will file close to 20 patents this year for companies in many spaces, including homeland security, anti-terrorism, several grid computing and virtual machine patents, and, ironically, I should have one issued in the expiring and disappearing e-mail arena. It was initiated 4-5 yrs ago.

Did he neglect to tell us about a patent for a system for software code comparison? When you read the patent description, you'll even find "spectral analysis" mentioned. Have we fingered the "MIT" deep divers at last? Anderer also revealed what he had thought was going to happen in the SCO litigation: "I think one real issue, that people are skirting, is who will be the ultimate guarantor of IP-related issues in a world that is governed by the GPL and GPL-like licenses. I could easily see IBM, HP, Sun, and many of the other large hardware players solving this problem tomorrow by settling the dispute with SCO and maybe even taking the entire code base and donating it into the public domain. I know this is what I originally thought would happen, at least the settlement part."

So sad, when dreams of wealth die, huh? Why would Sun, for one, pay anything for code it has already open sourced, very carefully in a way to keep the GPL out, and then put that code in the public domain? Why would IBM reward SCO's vicious behavior toward them? It probably sounded like a great strategy back in the planning stage to Darl and Mike, but look at it now, how it has played out.

Groklaw already covered the early days and the friendship between Anderer and McBride going back to the '90's, and their business adventures together before SCO began suing the world, and even then, we said that Anderer seemed more involved in the SCO strategy than at first appeared. Now, comes this patent application, which indicates even deeper involvement, if our guess that it's the same man proves to be accurate.

You might recognize some of the other inventors too, from Pointserve: Edward Powell and Mark Lane and also Ed White, who was "formerly co-founder and Vice-President of Engineering of PointServe". When IBM sent PointServe a subpoena, we wrote this:

You remember PointServe. Darl McBride was CEO at PointServe prior to joining FranklinCovey in 2000. And guess who is on the board at PointServe? McBride's old friend, Mike Anderer, the old pal who helped McBride come up with SCO's IP litigation strategy and approached Microsoft on SCO's behalf, according to Newsweek's Brad Stone, which led to the Baystar hookup.

Frank Sorenson noticed something else that is positively riveting. Check out PointServe's management page: G. Edward Powell, Chairman and Chief Executive Officer - ... For eight years prior to founding PointServe, Dr. Powell was a Member of the Technical Staff at the MIT’s Lincoln Laboratory

Mark T. Lane, Chief Scientist - ... Dr. Lane spent 10 years as a member of the Technical Staff at MIT Lincoln Laboratory

So, what do you think? Has IBM found the "MIT rocket scientists" Darl bragged about and then backed away from and that we imagined had somehow disappeared into the mist or into the Bermuda Triangle, thus becoming, alas, unavailable to testify for SCO as experts at trial about their purported comparisons of UNIX/Linux code?

Looks to me like we might have found not only the "MIT" rocket scientists SCO first bragged about and then clammed up about, but perhaps their method as well. Note some of what the patent says the invention can do:

[0034] For example, the present invention allows an owner of proprietary code to submit their code to a website which compares the code against a database of open source code bases. The database of open source code bases may be an open source UNIX or Linux distribution, for example. In this capacity, the system of the present invention would be used for auditing proprietary code to determine if it contained open source software, or if a particular open source software release contained proprietary software. This audit could be scheduled to run on a periodic basis automatically.

[0035] In another aspect, the database of open source code bases may contain a number popular open source applications of a certain type, such as image manipulation or audio processing applications that may be protected by trade secrets or patents. In this aspect, the input file may be patent claims or design specifications containing concepts that are compared against the concepts in the source code in the database. Thus, while the structure of the two corpuses is different (patent claims on the one hand and source code on the other) it is still possible according to the system of the present invention to determine whether they share concepts in common. . . .

[0038] For example, in another exemplary embodiment, the system of the present invention may take search phrases for comparison against a target corpus of patents, or patent claims for use in a patent search. As discussed earlier, the system parses the search terms for concepts based on natural language processing methods, and assigns raw power values based on the frequency of the concept in the target corpus. In a further aspect, the system may analyze each file in the target corpus (each patent in a patent database, for example), and replace each instance of each concept in the specifications with that concept's respective power. In addition, or in the alternative, the concept may be replaced by its power in the claims, in the case of an infringement analysis. . . .

[0043] More particularly, the process of profiling the corpus involves a multi-source characterization of that corpus along with a one-way transform assigned to preserve the confidentiality, secrecy and integrity of that original code document. Because the corpus may contain trade secrets or other proprietary intellectual property information, it may be necessary to use cryptographic methods to convert that corpus, which is readable by anyone, into a form that is only readable and useful by the system of the present invention and such that the conversion may not be reversed. This protects against the risk that the original corpus may be reverse engineered from the transformed corpus.. . .

[0047] However, a content analysis as noted in quadrant 102 can also be done to determine whether the content of two corpuses is similar even though their structure may differ. The content analysis may use rare word searches to accomplish this function. In the embodiment discussed earlier with respect to source code files and computer programming, while computer programming languages have certain reserved words that are likely to be found in any source code file written in that language, it is not likely that variable names, function names, procedure names or comments will be shared across source code files unless they were written by the same person or unless one was written with the knowledge of the other. Thus, the variable name, function name, procedure name or comment could be the rare word that is searched for in both corpuses. If the rare word is found in both, then it is likely that portions of source code were copied but simply altered in their structural position in the document. For example, if one code file uses an "if-then" statement and another corpus uses a "case" statement, but the variables are the same in the two code files, then the resemblance will be detected by the content analysis using rare word searches. This may reveal that the second corpus code file was written with the in the presence of or with the knowledge of first corpus, that the second corpus was written by someone who also wrote the first corpus, or that the second corpus is simply a rewrite of the first corpus.

[0048] Furthermore, and as illustrated in FIG. 1 at quadrant 103, while two corpuses may have the same structure, they may have different content. In this case, the system of the present invention may perform a spectral or histogram analysis to determine whether certain concepts are found in both documents despite being identified by different terms in the source code file. Thus, in the case of source code, structure could be an "if-then" statement used in both code files. However, if the two code files different variable names within this same structure, the resemblance will not be detected either by a strict textual analysis or content analysis using rare word searches. However, the spectral analysis 103 will detect the presence of similar structure where the rare words, in this case the variable names, are different.

Ah! SCO's favorite words: "concepts", as in methods and concepts. Get it? And "spectral analysis". I think X marks the spot, all right. And so handy for finding patent infringement just when Microsoft is allegedly wanting to find some. Is that a dovetail, or what?

Well, gang, you probably don't need me to tell you what I think the scheme was and is. But don't forget one salient truth: the examples SCO already offered, supposedly found by three teams of deep divers, including the "MIT" scientists, or... um, not *exactly* MIT scientists, or the used-to-be MIT-linked group, depending on which day SCO was blabbing, were shot down in a day:

In a slide show last Monday, SCO showed six examples of Linux files it says were illegally copied from its confidential Unix code.

Linux partisans who obtained a copy of the slide show were quick to trace the examples back to their origins, which appear preliminarily in each case not to belong to SCO.

The company disputes this analysis. “We’re the owners of the Unix (AT&T) System V code, and so we would know what it would look like,” the company told McMillan and the IDG News Service. “Until it comes to court, it’s going to be our word against theirs.”

Not exactly. Not only did the Linux community reveal the code wasn't useful for SCO's purposes, Judge Kimball also found nothing but a surprising lack of evidence. This patent may describe a system for code comparison, but it failed at the most important part, from SCO's standpoint: establishing who actually owns the code.

Prior art, anyone? Here's the patent claims and description, for those who would like to read the entire thing:

*****************************

United States Patent Application 20050216898
Kind Code A1
Powell, G. Edward JR. ; et al. September 29, 2005


System for software source code comparison

Abstract

A system for analyzing similarities between a first and second corpus or between a set of concepts and a corpus uses natural language processing and machine intelligence methods to replace terms or phrases in the corpus with concepts, determine the frequency of each concept in the corpus, and convert the corpus into a concept frequency file to enable easy comparison of the two corpuses or easy retrieval of items from the corpus that contain concept. Difference analysis and a combination of content and spectral analysis may be employed.


Inventors: Powell, G. Edward JR.; (Brentwood, TN) ; Anderer, Michael; (Salt Lake City, UT) ; Lane, Mark T.; (Franklin, TN) ; White, N. Edward; (Austin, TX)
Correspondence Name and Address:
    MCKENNA LONG & ALDRIDGE LLP
    1900 K STREET, NW
    WASHINGTON
    DC
    20006
    US
Serial No.: 938844
Series Code: 10
Filed: September 13, 2004

U.S. Current Class: 717/141; 717/114
U.S. Class at Publication: 717/141; 717/114
Intern'l Class: G06F 009/45


Claims


What is claimed is:

1. A system for comparing at least a first corpus to a second corpus, comprising: a profiler that characterizes each of said first corpus and second corpus; an encryption engine respectively encrypting the first corpus and the second corpus using a one-way transform; an analyzer identifying concepts in the transformed corpuses, said analyzer determining a frequency rating of said concepts in each corpus; for each corpus, replacing each instance of each of said concepts on every line with its respective frequency rating to create a frequency file; and a comparator correlating the frequency file for the first corpus to the frequency file for the second corpus.

2. The system of claim 1, wherein said profiler, encryption engine, analyzer, and comparator are computer programs running on at least one general purpose computer.

3. A system for searching a corpus of data objects, comprising: receiving a list of concepts; relating at least one of said concepts to at least one search term; searching each of said data objects for each of said terms; and determining the correlation of at least one concept and at least a second concept in said corpus of data objects based on the presence of search terms relating to said first and search terms relating to said second concept in the same data object.

4. The system of claim 3, wherein said corpus of data objects is a database of documents.

5. The system of claim 4, wherein said receiving a list of concepts comprises a computer program receiving a list of concepts posted from an internet web page.

6. The system of claim 3, said determining the correlation further comprises separately determining the correlation of each concept with each other concept.

7. The system of claim 3, said determining the correlation further comprises determining the correlation of each concept with three or more other concepts.
Description


[0001] This application claims the benefit of U.S. Provisional Patent Application No. 60/502,098, filed on Sep. 11, 2003, which is hereby incorporated by reference for all purposes as if fully set forth herein.

BACKGROUND OF THE INVENTION

[0002] 1. Field of the Invention

[0003] The present invention relates to data object comparison and analysis, and in particular to software for comparing two or more data objects to determine the extent of any similarities between them.

[0004] 2. Discussion of the Related Art

[0005] Companies increasingly rely on software to provide not only products for consumers or their institutions, but also to manage their day-to-day operations. Software code has therefore become a valuable intellectual property (IP) asset.

[0006] The ever-increasing complexity of computer software programs as well as tight development schedules force programmers to become more efficient. One way for programmers to meet these challenges is by reusing source code and adapting it to new applications rather than writing the source code from scratch.

[0007] To this end, open source software has become increasingly popular. Open source software is software source code that is publicly available and freely downloadable from the Internet. Thus, open source software code is a convenient resource for programmers looking to cut development time by downloading it and merging it with their proprietary application. In addition, the growth of the open source software movement may also motivate computer programmers to donate or contribute software to the open source movement that they have written but that is owned by their employer. The problem is that most open source software, while freely available for downloading is not in the public domain.

[0008] In particular, open source software is not unrestricted--to the contrary it is often subject to licenses that restrict not only the open source software code itself but any modification thereof and any software that incorporates it as well. Typically, these open source licenses may require that the source code of any proprietary system using some open source software code be publicly disclosed. In other words, a programmer who uses open source code in a proprietary application may unintentionally subject that proprietary application to the constraints and restrictions of an open source license. This may have devastating affects on the ability of the company to protect software IP or pursue further intellectual property protection for their software.

[0009] In addition, open source software has another inherent risk--it is unknown to what extent open source software incorporates proprietary technology owned by others. Thus, even if open source software is free of any licensing restriction, such as open source software that is in fact committed to the public domain, the possibility remains that the software may infringe another's patents or property rights. A programmer who incorporates this open source code into their proprietary application may unintentionally subject his employer to unforeseen consequences such as infringement litigation.

[0010] Furthermore, the rapid growth of the software industry has driven many programmers and software engineers to change employers regularly and often. There is a problem that as these workers move between jobs, they may be taking proprietary source code that they wrote for a previous employer with them to their new employment. Programmers may not be aware or may not be sensitive to these concerns, and risk an inadvertent technology transfer or intellectual property transfer.

[0011] In addition, as companies increasingly rely on overseas or offshore development firms for software programming, there is a concern that the overseas development company may be reusing source code that it wrote for one client (who has rights to that software) for projects it works on with other clients.

[0012] The problem is not limited to computer source code. In addition to source code, design documents and technical specifications may be indicative of patent infringement or may be used to invalidate patents. But due to the relative ambiguity of terms of art in the software and business methods fields as well as the non-technical nature of language that is often used in patents, it is very difficult to assess IP risks properly and efficiently.

[0013] These IP risks are more serious given the tight regulatory environment in which companies operate. Corporate regulations, such as those collectively known as "Sarbanes-Oxley", require that firms monitor their intellectual property assets as well as the financial risks to their business perform regular IP and risk audits, and report the same to their shareholders, regulators, and the public.

[0014] But given that programmers often modify source code slightly when reusing it, it becomes difficult to perform IP software risk audits using redline or other character-based comparison methods. Thus, what is needed in the art is a multi-dimensional approach to comparing two or more corpuses, such as source code, documents, file objects, collections of data or file objects, or databases, that is able to determine the extent to which one corpus resembles another even when the particular structure or content of the two corpuses vary.

SUMMARY OF THE INVENTION

[0015] Accordingly, the present invention is directed to a system for software source code comparison that substantially obviates one or more of the problems due to limitations and disadvantages of the related art.

[0016] An advantage of the present invention is to provide a system for comparing two corpuses to determine how they resemble one another.

[0017] Another advantage of the present invention is to provide a system, software, and methods for analyzing at least two corpuses and determining concepts contained in each and further determining the extent to which the corpuses contain concepts in common.

[0018] Additional features and advantages of the invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

[0019] To achieve these and other advantages and in accordance with the purpose of the present invention, as embodied and broadly described, a system for comparing at least a first corpus to a second corpus includes a profiler that characterizes each of said first corpus and second corpus; an encryption engine respectively encrypting the first corpus and the second corpus using a one-way transform; an analyzer identifying concepts in the transformed corpuses, said analyzer determining a frequency rating of said concepts in each corpus, replacing each instance of each of said concepts on every line with its respective frequency rating to create a frequency file; and a comparator comparing the frequency file for the first corpus to the frequency file for the second corpus.

[0020] In another aspect of the present invention, a system for searching a corpus of data objects includes: receiving a list of concepts; relating at least one of said concepts to at least one search term; searching each of said data objects for each of said terms; and determining the correlation of at least one concept and at least a second concept in said corpus of data objects based on the presence of search terms relating to said first and search terms relating to said second concept in the same data object.

[0021] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are intended to provide further explanation of the invention as claimed.

BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention.

[0023] In the drawings:

[0024] FIG. 1 is a diagram illustrating an aspect of a first exemplary embodiment of the present invention.

[0025] FIG. 2A is a process diagram illustrating the system of the present invention according to a first exemplary embodiment.

[0026] FIG. 2B is a process diagram illustrating profiling according to a first exemplary embodiment of the present invention.

[0027] FIG. 3A illustrates sample histograms according to an aspect of a first exemplary embodiment of the present invention.

[0028] FIG. 3B illustrates sample spectral extracts according to an aspect of a first exemplary embodiment of the present invention.

[0029] FIG. 4 illustrates a sample correlation matrix according to an aspect of a first exemplary embodiment of the present invention.

[0030] FIG. 5 illustrates a further embodiment of the present invention.

DETAILED DESCRIPTION OF THE ILLUSTRATED EMBODIMENTS

[0031] Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings.

[0032] The system of the present invention models the conditional probability that two (or more) corpuses have a similar combination of characteristics. For example, the two corpuses may be software source code bases composed of source code files, structured or unstructured documents, patents, or technical disclosures. The characteristics analyzed may be the structure and content of those code bases and source code files, for example.

[0033] The system of the present invention analyzes and compares the corpuses in such a way that they may be preprocessed without affecting the comparison. In one exemplary embodiment, the corpus is transformed using any one of a number of one-way transforms understood to those of ordinary skill in the art, allowing the system of the present invention to protect the proprietary, secure, confidential, or privileged nature of the corpus and still allow it to be compared against another corpus. In the alternative, proprietary one-way encryption transforms may be used.

[0034] For example, the present invention allows an owner of proprietary code to submit their code to a website which compares the code against a database of open source code bases. The database of open source code bases may be an open source UNIX or Linux distribution, for example. In this capacity, the system of the present invention would be used for auditing proprietary code to determine if it contained open source software, or if a particular open source software release contained proprietary software. This audit could be scheduled to run on a periodic basis automatically.

[0035] In another aspect, the database of open source code bases may contain a number popular open source applications of a certain type, such as image manipulation or audio processing applications that may be protected by trade secrets or patents. In this aspect, the input file may be patent claims or design specifications containing concepts that are compared against the concepts in the source code in the database. Thus, while the structure of the two corpuses is different (patent claims on the one hand and source code on the other) it is still possible according to the system of the present invention to determine whether they share concepts in common.

[0036] In these aspects of the invention, the need to keep the proprietary corpus confidential is paramount. Thus, providing a one-way transform of the proprietary corpus, using some form of or combination of natural language processing, machine learning, and data encryption, minimizes the risk of inadvertent disclosure of proprietary information. It is necessary that the transform be one way (i.e. irreversible) to protect the confidentiality of the corpus against the risk that the system on which the comparison is run is compromised in some way, or that the corpus is intercept en route.

[0037] As noted earlier, the system is not limited to comparing source code. The system may be adapted to compare compiled object code as well, which is important in case of reverse engineering or infringement of copyrights to software. Furthermore, the system may be adapted to corpuses other than computer code.

[0038] For example, in another exemplary embodiment, the system of the present invention may take search phrases for comparison against a target corpus of patents, or patent claims for use in a patent search. As discussed earlier, the system parses the search terms for concepts based on natural language processing methods, and assigns raw power values based on the frequency of the concept in the target corpus. In a further aspect, the system may analyze each file in the target corpus (each patent in a patent database, for example), and replace each instance of each concept in the specifications with that concept's respective power. In addition, or in the alternative, the concept may be replaced by its power in the claims, in the case of an infringement analysis.

[0039] In a first exemplary embodiment the present invention allows one corpus to be compared against at least one other corpus. As noted, a corpus may be any data object, file object, collection of data or file objects or any type of structured or unstructured data or documents. This includes source code files including both instructions and comments, object code, text documents, structure documents such as spreadsheets, word processing files, HTML or XML documents, or databases or collections thereof.

[0040] In a first aspect of this invention, a first corpus (the source corpus) is profiled and converted into a metadata file. Likewise, the second, or target, corpus is profiled and converted into a target metadata file. In this particular aspect, the profiling process includes encrypting or otherwise transforming the corpus using a one-way transform, and then characterizing the transformed corpus before converting it into a metadata file.

[0041] FIG. 2A is a block diagram generally illustrating a system for comparing two corpuses according to the present invention. Proprietary intellectual property is taken at step 20 as input and transformed at step 22 using natural language processing, machine intelligence, and encryption. At step 24, the transformed proprietary property is characterized as discussed herein and compared at step 26 with one or more other characterized corpuses in the characterization database 28. The profiling tool may perform multi-source characterizing and a one-way transform. By making the transform a one-way transform the system will protect the proprietary nature of the source code. If the source code could be reverse engineered from the metadata file, very few companies with proprietary source code would want to use the system for fear of disclosing their source code to others. By making it a one-way transform, they may be comfortable that their confidential information and source code will be kept confidential.

[0042] Software source code B is taken as input at step 21 by a profiling block 23 which performs profiling on the source code to produce a metadata file B at step 25. The metadata files are then compared to one another at step 26 and a report is generated at step 27. The report will reflect how closely the two pieces of software resembled one another.

[0043] More particularly, the process of profiling the corpus involves a multi-source characterization of that corpus along with a one-way transform assigned to preserve the confidentiality, secrecy and integrity of that original code document. Because the corpus may contain trade secrets or other proprietary intellectual property information, it may be necessary to use cryptographic methods to convert that corpus, which is readable by anyone, into a form that is only readable and useful by the system of the present invention and such that the conversion may not be reversed. This protects against the risk that the original corpus may be reverse engineered from the transformed corpus.

[0044] After the two corpuses are profiled and converted into respective source and target metadata files, then the two metadata files are compared to determine how closely they resemble each other. The details of the multi-source characterization and the comparison will be discussed below.

[0045] In further aspect of the first exemplary embodiment, the corpus is characterized by a structure and content. In other words, any data object or file object will contain some inherent structure that organizes the content stored within it. Thus, it is possible for two corpuses such as source code files, for example, to have both similar structure and content, different structure and content, similar structure with different content, or different structure with similar content. A two-by-two matrix showing the possible scenarios is illustrated in FIG. 1.

[0046] FIG. 1 illustrates the possible relationships between two corpuses. The system of the present invention can perform a number of different analyses on two corpuses to determine whether they resemble one another. For example, to determine whether the two corpuses share content and structure as noted in quadrant 101, ordinary text comparison programs such as redline applications or text comparison commands, such as the grep, diff or comm commands found in the UNIX operating system, may be used. This will reveal whether or not sections of the corpus are identical in structure and content.

[0047] However, a content analysis as noted in quadrant 102 can also be done to determine whether the content of two corpuses is similar even though their structure may differ. The content analysis may use rare word searches to accomplish this function. In the embodiment discussed earlier with respect to source code files and computer programming, while computer programming languages have certain reserved words that are likely to be found in any source code file written in that language, it is not likely that variable names, function names, procedure names or comments will be shared across source code files unless they were written by the same person or unless one was written with the knowledge of the other. Thus, the variable name, function name, procedure name or comment could be the rare word that is searched for in both corpuses. If the rare word is found in both, then it is likely that portions of source code were copied but simply altered in their structural position in the document. For example, if one code file uses an "if-then" statement and another corpus uses a "case" statement, but the variables are the same in the two code files, then the resemblance will be detected by the content analysis using rare word searches. This may reveal that the second corpus code file was written with the in the presence of or with the knowledge of first corpus, that the second corpus was written by someone who also wrote the first corpus, or that the second corpus is simply a rewrite of the first corpus.

[0048] Furthermore, and as illustrated in FIG. 1 at quadrant 103, while two corpuses may have the same structure, they may have different content. In this case, the system of the present invention may perform a spectral or histogram analysis to determine whether certain concepts are found in both documents despite being identified by different terms in the source code file. Thus, in the case of source code, structure could be an "if-then" statement used in both code files. However, if the two code files different variable names within this same structure, the resemblance will not be detected either by a strict textual analysis or content analysis using rare word searches. However, the spectral analysis 103 will detect the presence of similar structure where the rare words, in this case the variable names, are different.

[0049] Finally, there may be instances that fall into the fourth quadrant 104 of FIG. 1, where both the structure of the document and the content of the document are different. This is where it is necessary to provide human IP or intellectual property thread analysis. In other words, human readable documents such as manuals, read-me files, message board postings, news group postings, chat transcripts, resumes, press releases, journal articles, and marketing materials or the like are reviewed to determine whether people involved in creating the first corpus were at a different time with working with the company that wrote the second corpus. In the alternative, such documentary analysis may reveal that authors of the first and second corpus knew each other, were familiar with one another, or those working or somehow came in contact with each other.

[0050] The spectral analysis will now be discussed in detail. In an aspect of the first exemplary embodiment, the corpuses being compared are source code. To characterize source code according to the present invention, each file in the code base is processed as illustrated in FIG. 2B. The processing involves stripping away any comments, white spaces or programming language-specific characters, such as the asterisk, the ampersand, semicolon, comma, for example, in step 202. It is understood by one of ordinary skill in the art that a different type of corpus such as a text document, XML document or HTML document will have different characters that are specific delimiters in that type of corpus.

[0051] After this information has been removed, at step 204, concept information is gathered from the source code files in the code base corpus. Concept information is gathered by first producing a raw concept file at step 206 which retains the line structure and that records the concepts in those lines in a dictionary file. Next, the raw power of each concept is determined at step 208. The raw power is the number of times that the concept is used in the entire code base.

[0052] After the raw power of each concept in the code base is determined, a raw concept frequency file for each source file in the code base is produced at step 210. This raw concept frequency file records the concepts on each line of the file by replacing the concepts on the line with their respective raw power values. After step 210, the system of the present invention according to this particular exemplary embodiment assigns a frequency or power number to every term used in the code file at step 212.

[0053] Thus, for each line in the file each concept is translated into the power of that concept from the corpus dictionary that was created earlier. For example, a line containing a number of different concepts would be replaced by a sequance such as 2363:12:300:41, for example, in which the numbers are the power numbers of the concepts and the colons are delimiters used to separate different concepts on the same line.

[0054] After this stage, spectral summary charts may be created as illustrated in FIG. 3A. The spectral summary chart reports on the similarities between the two code files A and B by providing graphs 301 and 302 of the histograms or spectrum of each of code files A and B, respectively, based the name of the file, the number of files and number of lines in the file, the number of distinct concepts used in the file and the total power of the lines in the file. This can then be plotted and displayed in an ordinary bar chart format as illustrated in FIG. 3 in which the horizontal axis is the line number of the file and the vertical axis is the total aggregate power of that line from the concept dictionary. By looking at the or spectral charts of the two files being compared, one can see immediately whether or not the files contain similar concepts because each line in each file will be replaced by a bar on which the concept values in that line are plotted. The similarities between the two files become obvious.

[0055] Furthermore, as illustrated in FIG. 3B, a spectral extract can be obtained in which portions of a histogram from one file can be compared against the histogram of the other file to see if there are sections of the histograms that match exactly. This can be used to determine whether or not entire sections of source code were duplicated in concept if not in precise exact character matching. In other words, because source code that accomplishes the exact same thing can be written in different ways, it is necessary to determine to what extent the source code is written using the same variable names, the same functions and the same order or using the same programming styles which under ordinary circumstances would differ significantly from one programmer to another. Thus, if sections of the code display similar identical concepts, it is very likely that source code has been duplicated and only modified slightly.

[0056] In a further embodiment of the present invention, the content analysis and spectral analysis may be further extended to analyze patents and patent claims for invalidity or infringement purposes. In other words, while an intellectual property document such as a patent or a design specification may include terms used to convey a concept, it is understood that there are other terms that may be used as synonyms for that same concept. This is particularly the case in software and business method patents where there are few industry standard terms of art, or in which the terms of art have ambiguous meanings and are used loosely by those in the art.

[0057] Thus, the system for the present invention may have at its disposal a corpus dictionary that is either predefined for a specific field of knowledge in which the corpus (the patent, in this example) resides or it may have a dictionary that is constructed ad hoc as part of the analytical process using the first and second corpus to produce the corpus dictionary of key concepts.

[0058] In addition, the concepts may be used to determine the extent to which concepts are highly correlated in a corpus. Consider an example in which the correlation of a number of biomedical concepts in patents is sought. In this example, ten concepts 400, "neuromodulation", "brain imag*", "cord stimulat*", "nerve stimulat*", "vivo magnetic resonance", "Interventional Magnetic Resonance or Interventional MR", "brain stimulat*", "intralaminar nucle*", "sympathetic or parasympathetic", and "corpus callosum", are entered into system of the present invention. (The * denotes a wildcard operator). The system, using natural language processing methods understood in the art, searches a set of patents or all patents for instances of the concepts (using terms from the concept dictionary synonymous with the concept). The system returns a grid such as that illustrated in FIG. 4, with the concepts 401 listed vertically along the left side and the correlated concepts 402 listed along the top. The number of patents 403 found containing each concept is returned and displayed along with the concept 401 at the left. Then the system correlates each concept with each of the other concepts and displays as a percentage 404 of the total patents 403 found containing the first concept alone the number of patents containing both search terms together. If implemented as a hypertext document or world-wide-web page, the percentage 404 can be selected to reveal the list of the patents having the respective concepts.

[0059] This embodiment is not limited to a two dimensional grid. In alternative aspects of this embodiment, a multidimensional array N1.times.N2.times. . . . .times.Ni returns the correlation of any of concepts 1 though i with any number greater than or equal to two of the other concepts 1 though i. Conceptually, a 10.times.10.times.10 cube would store the correlations of three of the ten concepts listed above. It will be understood to those of skill in the art at the time of the invention that the system of the present invention may be implemented in any number of ways.

[0060] For example and as illustrated in FIG. 5, the present invention may be implemented as a site on the internet which aggregates publicly available documents on the internet, such as source code or patents, onto databases residing on its own system which are used for the comparison. In another example, the present invention may periodically access open source code bases or patent databases across the internet and compare them against proprietary code that is stored on its databases and servers to provide periodic IP monitoring and auditing.

[0061] It will be apparent to those skilled in the art that various modifications and variation can be made in the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention cover the modifications and variations of this invention provided they come within the scope of the appended claims and their equivalents.

  


Mike Anderer, "MIT" Scientists, Spectral Analysis, and a Patent Application | 203 comments | Create New Account
Comments belong to whoever posts them. Please notify us of inappropriate comments.
OT Here, Please
Authored by: TheBlueSkyRanger on Friday, September 30 2005 @ 04:16 AM EDT
Please remember to keep links clickable.


Dobre utka,
The Blue Sky Ranger

[ Reply to This | # ]

Corrections Here, Please
Authored by: TheBlueSkyRanger on Friday, September 30 2005 @ 04:17 AM EDT
If needed.

Dobre utka,
The Blue Sky Ranger

[ Reply to This | # ]

loony assertion no. 19456: most open source code stolen from employers?
Authored by: Anonymous on Friday, September 30 2005 @ 04:54 AM EDT

Does this kind of sweeping generalisation...

[0007] ...open source software movement may also motivate computer programmers to donate or contribute software to the open source movement that they have written but that is owned by their employer. The problem is that most open source software, while freely available for downloading is not in the public domain.

... really tick you off?

If a statement that is not clearly supported by fact, reference or example is included in a patent, does it go some way to diluting the potential for the patent to be accepted?

It clearly displays the alleged inventors' bias against the open source movement

It also shows to me what a farce the software patent thing is - no doubt the system to do this hasn't beeen coded and proved to actually function - and IBM have probably done better with the own code comparison experts - so correct me if i'm wrong, but this is nothing more than a patent application for vapourware?

Charles from Oz

[ Reply to This | # ]

"MIT" Scientists humor...
Authored by: Nonad on Friday, September 30 2005 @ 05:00 AM EDT
Sorry, I couldn't resist popping this in here again...

Red Meat take off...

[ Reply to This | # ]

Lemelson-style "invention"
Authored by: Totosplatz on Friday, September 30 2005 @ 05:03 AM EDT

Blue-sky rumination Lemelson-style "invention" - toss in every kind of idea plus the kitchen sink...

From now on any kind of system that compares something to some other thing in a repository somewhere will "infringe" the patent on this "invention." With this description in hand, can one of these beasts actually be implemented? "Spectral" analysis on the 'concepts' present? In what exact century? After encryption has rendered it all into mush? Give me a break!

Toss out the USPTO! Clearly not up to the task. Either that or this is some kind of grad-student prank!

---
All the best to one and all.

[ Reply to This | # ]

Previous Art
Authored by: Anonymous on Friday, September 30 2005 @ 05:15 AM EDT
The process of profiling the corpus involves a multi-source characterization of that corpus along with a one-way transform assigned to preserve the confidentiality, secrecy and integrity of that original code document.

For previous art on this, how about Eric Raymond's comparator program, or my own comparison tool. Both of them perform a one-way transform on source code, and both were published well before September 13, 2004.

Warren

[ Reply to This | # ]

Does this patent have any merit, or just apllied to give the projected FUD credability?
Authored by: Anonymous on Friday, September 30 2005 @ 05:24 AM EDT
Could anybody comment on whether this patent has any merit?

From reading it I get the impression that it is complete hogwash, pseudo
scientific rambling. As such it reminds me of the Alan Sokal hoax (use google).

Several things I find rather suspect:

The text speaks of comparing UNIX source code to Open Source code. Funny that
the wording is tailored to newSCO's needs. Why, if one has a method for
comparing UNIX source code to Open Source code, could this method not be apllied
to any two pieces of source code? Why on earth would one not make the patent as
broad as possible?

The problem of deciding whether two pieces of code have the same semantics, ie,
perform the same task, is undecidable: no procedure can exist that within a
finite amount of time tells one whether or not the two pieces of code have the
same sematics. And this patent application pretends to be able to do even
better: whether one piece of code implementens a patent or not.

Spectral analasys an source code comparison don't go hand in hand. This patent
does not tell me how they could. For finding structural similarities one would
rather apply pretty straighforward technique: compile the code to a suitable
representation (a tree), optimize, and than compare.

I got the distinct feeling that this patent is applied just to unerpin the FUD
("We used patented technology to establish blah blah infringment blah blah
Linux blah blah patents blah blah <more McBride & Stowell speak
here>)



[ Reply to This | # ]

stats_for_all should apply for a patent
Authored by: caliboss on Friday, September 30 2005 @ 05:41 AM EDT
Seriously...Mr. Stats_For_all is one fellow I would never want on the other side of a dispute.

Is there some secret sauce here that you use or can you give us all some insights, tips, tricks, cluesticks as to how you soooo effectively eviscerate SCO before they even have a chance to exhale.

What's your research setup look like? What data sources do you use and trust? I'm amazed.

---
Grok the Law / Rock the World

[ Reply to This | # ]

Ultimate Guarantor
Authored by: inode_buddha on Friday, September 30 2005 @ 05:42 AM EDT
Memo to Mike Anderer: The "ultimate guarantor" already exists in the
form of 17 USC. It doesn't need any help from anyone.

---
-inode_buddha
Copyright info in bio

"When we speak of free software,
we are referring to freedom, not price"
-- Richard M. Stallman

[ Reply to This | # ]

Prior Art? (From ESR)
Authored by: eamacnaghten on Friday, September 30 2005 @ 05:56 AM EDT
Prior art, anyone?

How similar is ESR's Comparator? Released (therefore published) under GPL in 2003 I think...

Link.

Web Sig: Eddy Currents

[ Reply to This | # ]

Hope it complies with the GPL
Authored by: thorpie on Friday, September 30 2005 @ 05:56 AM EDT

Now if they are using open source source code in their program, even as an input, do they have to comply with the GPL??

I'd really like to know

---
The memories of a man in his old age are the deeds of a man in his prime - Floyd, Pink

[ Reply to This | # ]

Prior Art - Natural Language Processing
Authored by: Anonymous on Friday, September 30 2005 @ 06:29 AM EDT
If I understand this correctly that this is about corpus analysis then we are in
the realms of "natural language processing" - which, among many other
things - is computational analysis of large textual works to glean meaning - new
words for dictionaries, eg., OED, are identified in this way.

US Groklawers could track outputs from this discipline via DARPA and a body
whose name escapes me (National Science Foundation? - that does all the science
funding)

EU Groklawers could track down work from the Commission's Telematics programme.
Universities and companies have done a lot of government funded work here.

UK Groklawers could check with the Research Councils or possibly the Department
of Trade and Industry - including asking about old annual reports of the JFIT
programme and or SALT.

For those of you in Universities - nip along to your (e.g., English) Language
departments, rather than Computational Sciences.

UK Groklawers in Universities could try English Language departments in
Imperial, Oxford, UCL, Birmingham, Lancaster, Essex, UCE, Edinburgh

HTH


[ Reply to This | # ]

Prior art?
Authored by: cmc on Friday, September 30 2005 @ 06:54 AM EDT
I seem to recall at least two companies popping up recently that claim to do code comparison (compare submitted code against databases of known code), specifically looking for IP infringement. I just did some googling, and one of those companies is Black Duck Software. Here's an article about their service in eWeek. From that article:

"Black Duck Software is now providing its source-code checking software as an Internet-based service. The software, ProtexIP, is used to check for any potential violations of open-source or other software licenses within source code. Now, customers can subscribe to Black Duck's knowledge base through its ProtexIP/OnDemand service rather than installing their own servers on-site."

This Infoworld article (dated May 14, 2004) mentions the launch of the ProtexIP software and what it does (code comparison). This is the packaged product, available long before the web service was launched, and even before Mr. Anderer's patent application was filed.

Might this ProtexIP be prior art, or are the two unrelated (legally-speaking)? Specifically, does it look like ProtexIP (and others that do the same thing) are infringing this Anderer patent?

cmc

[ Reply to This | # ]

  • Prior art? - Authored by: Anonymous on Friday, September 30 2005 @ 09:00 AM EDT
  • Prior art? - Authored by: Anonymous on Friday, September 30 2005 @ 02:25 PM EDT
Mike Anderer, "MIT" Scientists, Spectral Analysis, and a Patent Application
Authored by: Anonymous on Friday, September 30 2005 @ 06:59 AM EDT
Uhm... Blackduck Software Doesn't this count as prior art? Or are they connected somehow?

[ Reply to This | # ]

Patent rubbish
Authored by: alextangent on Friday, September 30 2005 @ 07:44 AM EDT

This patent is complete rubbish; the method outlined for comparison is not described in any detail and is not implementable in any universe of which I am a resident. For instance, talking about the protection of the target for comparison, the patent states

[0036] In these aspects of the invention, the need to keep the proprietary corpus confidential is paramount. Thus, providing a one- way transform of the proprietary corpus, using some form of or combination of natural language processing, machine learning, and data encryption, minimizes the risk of inadvertent disclosure of proprietary information. It is necessary that the transform be one way (i.e. irreversible) to protect the confidentiality of the corpus against the risk that the system on which the comparison is run is compromised in some way, or that the corpus is intercept en route.

How is the comparsion to be performed if the target is irreversibly encrypted?

[0037] As noted earlier, the system is not limited to comparing source code. The system may be adapted to compare compiled object code as well, which is important in case of reverse engineering or infringement of copyrights to software. Furthermore, the system may be adapted to corpuses other than computer code.

Really? Do these guys understand reverse engineering or compiler technology at all?

[0053] Thus, for each line in the file each concept is translated into the power of that concept from the corpus dictionary that was created earlier. For example, a line containing a number of different concepts would be replaced by a sequance [sic] such as 2363:12:300:41, for example, in which the numbers are the power numbers of the concepts and the colons are delimiters used to separate different concepts on the same line.

I had to stop there. The rest is nothing more than psuedo-computing mumbo jumbo and a lot of hand waving. Perhaps someone else has the patience to dissect this nonsense; I don't. Unfortunately, it appears the patent examiner didn't have the patience either.

[ Reply to This | # ]

Mike Anderer, "MIT" Scientists, Spectral Analysis, and a Patent Application
Authored by: LouS on Friday, September 30 2005 @ 07:46 AM EDT
The idea of comparing texts by taking histograms of word usage is a very old one in natural language information retrieval. Just for one example see Lewis, D. D.; Yang, Y.; Rose, T.; and Li, F. RCV1: A New Benchmark Collection for Text Categorization Research. Journal of Machine Learning Research, 5:361-397, 2004. http://www.jmlr .org/papers/volume5/lewis04a/ lewis04a.pdf 

[ Reply to This | # ]

Patenting the evidence gathering process
Authored by: Anonymous on Friday, September 30 2005 @ 08:04 AM EDT

Does the process of discovery and code comparison which took place in the SCO-IBM litigation constitute prior art?

If this patent is actually granted, will this not limit the ability of copyright holders of Free/Open source software to prove a claim of infringement by non-free software vendors? -- Because evidence in such litigation will involve comparison of Free/Open source code with proprietary source code?

Is the latter real intention behind filing this patent?

Does comparison of closed and open sourced code involve any kind of innovation?

Is not code comparison already done by Free software utilities like diff?

[ Reply to This | # ]

Mike Anderer, "MIT" Scientists, Spectral Analysis, and a Patent Application
Authored by: Anonymous on Friday, September 30 2005 @ 08:12 AM EDT
What a bunch of do-nothing get rich quick paper shufflers.

[ Reply to This | # ]

Lets patent everything!
Authored by: Anonymous on Friday, September 30 2005 @ 08:15 AM EDT
If this patent can get through the system then it's time to patent everything!

Does not matter if it works, just write it down!

Let's see...

I'll patent an invention that turns water into wine. I don't need to explain the proccess, just state a few quick claims and viola! Instant patent!

What this all boils down to is a patant overhaul. All companies have a choice between "trade secrets" and "patents". Take Coke as and example, they chose to keep the formula for their tasty beverage as a trade secret, if they had gone the patent route the forumla would be common knowledge. As soon as you file the patent the information on how to replicate/create what your are applying a patent for should be in the patent.

You can't just write down some jargon and say, "I just patented cold fussion!" You must state, in easily understandable instructions, how your patent actually works. Not what it does, but the nuts and bolts of how it does what it does. That is the trade-off when you patent as opposed to when you keep something as a trade secret.

For those who say, "I have to patent or I lose my legal ownership." Hogwash! If someone acquired the forumla for Coke and tried to patent it tomorow they would not get it as there is prior art available at your local store! The formula would lose it's trade secret status, but that's about it. (well except for a bunch of lawsuits and finger-pointing.)

I propose the patent office go back to the way it used to be with complete drawings/plans/schematics required for ALL applications. If you can not show the patent office EXACTLY how to make the patent work (I'm not talking about a slideshow but hard evidence/prototype/source code) then the application should be rejected out of hand.

Without such a requirement all patents are a joke.

[ Reply to This | # ]

Mike Anderer, "MIT" Scientists, Spectral Analysis, and a Patent Application
Authored by: Anonymous on Friday, September 30 2005 @ 08:44 AM EDT
First thanks to stats for this gem.

Now for the other readers here I want you to cast your minds back in time.

First we have the MIT scientists, spectral analysis and code comparison.

We have references here to one way encryption to protect the closed source code.


We have a buddy of D McB behind this patent. Not first named now which might or
might not be by itself significant here.

The patent makes extensive reference to Linux and Unix.

The filing date is interesting also - 2004.

Lets make a guess at dot connection here.

Back to 2002. Yarro sees that buying the Unix divisions isnt helping. He comes
up with plan B - to use the control of Unix code to initiate a law suit against
Linux. He needs a mouth pice to sell this idea: it is never intended to go to
court note - just create a stink to get the company out of Canopy so he can make
yet another land grab for Canopy.

D McB is approached. He rings up his friend Anderer. Anderer figures that he can
code something with a few friends that will pull up a few code similarities
between Unix and Linux. There will be +ves - grep, chron etc - the question is
how many?

D McB goes back to yarro and says: "Yes we may have something here. We
[Anderer at al] develop a code comparion system that will pick out similarities
between Unix and Linux for you. I [D McB] will sell this idea for you. I return
I get to hire my buddies."

The deal is done.

The one way encryption is the Greek text stuff (remember that?). The spectral
analysis is applied to the histogrames of the word counts. (This is a genuine
technique but its interpretation is often tricky at best.)

The missing money that stats also found probably relates to this use of this new
idea by MS. MS wanted this idea for the FUD machine. The SCO deal was the first
go at making this work. If this had worked this patented idea would not be in
use on a widescale by MS to create at least more FUD.

Now we may have found why Anderer though they might be able to get more money
out of MS. If for example OO could be shown to share code with (say) MS Office
OO would be now deep in the weeds - assuming that this patent held up in court
etc. This is why the parent was never assigned to SCO: SCO was to be bought out
by IBM and the patent would have gone with it.

My guess on why Anderer is not the first named on the patent is because of his
visibility in this case. It might also relate to his contribution of course to
the patent but Im just being 'paranoid' here.

The results of this method would be convincing to a non techie - think D McB,
Laura Didot, possible M' OG & Enderle - who knows who say the results of
this programme of Anderer's. Needless to say it would pull out sections of
similar code: chron, grep, the interfaces etc which would be most impressive. So
impressive that Yarro hired D McB in fact.

D McB is no techie. Yarro is no techie. Neither are copyright lawyers. The ex
MIT CVs would look good. The mats qualification look ood. They run this past the
in house lawyers. They decide to run with it starting out with Kevin McBride (D
McB brother) in a state court where they think they have a better chance of
sucess.

Then it goes pear shaped for them. IBM want to fight. The case gets transfered.
Instead of being a cheap and easy case they have to hire expensive lawyers
(BSF). The evidence is adequate at first blush for BSF to think there might just
be a case. SCO's methods wont stand up in court so they are forced to try
manual comparisons.

MS are watching the case carefully. If the method had stood up in court or if
IBM had bought out SCO there would now be mountains of claims based on this
patent which alsmot certainly would have been bought by MS for thier own use
agaist IBM and anyone else that had dared to touch FOSS code.

This by itself IMHO justifies IBM's decison to fight this case all the way.
Otherwise they would have been sitting on a lot more trouble in the future.

--

MadScientist

[ Reply to This | # ]

Excuse Me?
Authored by: rsteinmetz70112 on Friday, September 30 2005 @ 09:11 AM EDT
This patent application seems specifically designed to buttress SCOG's claims
and to provide a basis for getting the inventors recognized as experts in the
field of software comparison. In other words the patent claims seem designed
specifically for this litigation.

I picked out this gem "it is not likely that variable names, function
names, procedure names or comments will be shared across source code files
unless they were written by the same person or unless one was written with the
knowledge of the other." This seems almost directly from SCOG's claims. It
also curiously omits the possibility of published standards, common practice,
industry conventions or shared (but not copied) experience (say everyone went to
MIT).

If these guys testify in the IBM case, I can't wait for IBM to ask for the
source code so their experts can determine the validity of the method.

---
Rsteinmetz - IANAL therefore my opinions are illegal.

"I could be wrong now, but I don't think so."
Randy Newman - The Title Theme from Monk

[ Reply to This | # ]

All well and good, but...
Authored by: rand on Friday, September 30 2005 @ 09:42 AM EDT
..."proving" that two things are similar doesn't prove that B was copied from A.

1. If you encrypt the data properly (squint and wear colored glasses), oranges look identical to apples.

2. In the case of source code, it's most probable that both A and B were copied from some prior source (Adm. Hopper and Knuth comes to mind). The invention might work, after a fashion, but to use it to demonstrate that a certain piece of open-source software was copied from proprietary* code would first require that the proprietary* code be compared to every piece of software previously written by anyone, anywhere, in any language.

*Again we see the usage of "proprietary" and "IP"/"intellectual property".
Remember, "proprietary" just means someone owns or claims something, as opposed to "public domain"; in that regard most open-source software is still proprietary (otherwise, GPL wouldn't work!).
Remember also, there is no intellectual "property", there are only limited rights: the right to copy (copyright), the right of exclusive manufacture (patent), the right of exclusive identification (trademark), and the right to protect trade secrets.

---
The wise man is not embarrassed or angered by lies, only disappointed. (IANAL and so forth and so on)

[ Reply to This | # ]

Filed _JUST_ after Eric Raymond's tool (that did this) released
Authored by: dwheeler on Friday, September 30 2005 @ 10:05 AM EDT
I have not had time to read this patent application, but I did look for the original filing date. At the top it says 2004; it then references an earlier filing of Sep. 11, 2003.

That is a strikingly interesting date. You see, Eric Raymond released his comparitor tool to the world on Sep. 7, 2003. See this news file. His comparator tool seems to be prior art, so this patent is probably invalid just on that basis. Indeed, I know that Raymond's is not the first tool to do this (there are others far older), so this patent may be DOA.

Speculation: It may very well be that this patent-applier read about Raymond's tool, and then quickly wrote a patent application to cover what his tool did after the tool had been publicly released to the world. If that is what happened, this isn't just invalid, it's unethical.

[ Reply to This | # ]

Bayesian
Authored by: Anonymous on Friday, September 30 2005 @ 10:10 AM EDT
Okay, I'll admit I didn't read the entire patent application...but, it sure
sounds like he's just taken bayesian logic to tokenize a submission and compare
it against a database of known previously tokenized data just like bayesian spam
filters do...only on code instead of e-mail.

[ Reply to This | # ]

Mike Anderer, "MIT" Scientists, Spectral Analysis, and a Patent Application
Authored by: John Hasler on Friday, September 30 2005 @ 10:25 AM EDT
Filed: September 13, 2004
[0001] This application claims the benefit of U.S. Provisional Patent Application No. 60/502,098, filed on Sep. 11, 2003...
I thought provisional patents were only good for one year.

[ Reply to This | # ]

New Tool To Scam Investors..
Authored by: Anonymous on Friday, September 30 2005 @ 10:47 AM EDT
This reminds me of some past scams where somebody gets a patent on
"something" and then uses the patent to attract investors, stock
analyists, bankers, and the even less informed public.

It is a "magic asset" that is used to inflate the value of a company
that is otherwise worthless. Think perpetual motion machine.

The bsst part is that it never really needs to work -- just give the appearance
of working.

[ Reply to This | # ]

IBM/Novell should talk to stats
Authored by: Anonymous on Friday, September 30 2005 @ 11:26 AM EDT

Stats has an amazing ability to parse large amounts of information and pull out
really interesting bits.

If I were on IBM or Novell legal team I would seriously think about bringing
Stats in, perhaps as a contractor and let him see all the real dirt on SCO (from
discovery and subpoenas) and not just information available from public sources
alone.

And then when its all over, he could write a book with all the gory details and
I'll be the first in line to get it.

[ Reply to This | # ]

Mike Anderer, "MIT" Scientists, Spectral Analysis, and a Patent Application
Authored by: Anonymous on Friday, September 30 2005 @ 12:52 PM EDT
I don't really care about the patent aspect of this (getting US patents seems
easier than applying for a credit card these days) but the business model that
they seem to be promoting (and which could be used even if the patent were
rejected).

Basically they seem to be building a service where anyone with a few lines of
source code will be able to submit it and have it compared to the entire bidy of
OSS work to find any 'matches'. They'll then get a report of those 'matches' and
head off down to the local court house and try playing the SCO game. Whats the
bet that you could put nearly any code at all in there and it would come up with
a 'match' of some kind. Seems like it's tailor made to create a storm of
nuisance law suites against OSS devs / distros etc. Perhaps it could even kill
some projects or distros by bogging them down in legal battles.

[ Reply to This | # ]

Trial by Jury
Authored by: mac586 on Friday, September 30 2005 @ 12:54 PM EDT
Another exciting piece of evidence that reveals the innerworkings of SCO's
"Dream Team".

When Darl and company commented on how much they desired a courtroom and a jury,
they never imagined that such a trial was going to take place online, in a
blog.

How can SCO even imagine a day in court? These days, they must be imagining
years in a federal prison.

[ Reply to This | # ]

Wow!
Authored by: gtall on Friday, September 30 2005 @ 01:10 PM EDT
He replacing "terms or phrases in the corpus with concepts".

So, how would this work for, say, "i = 0"?

(1) the term "i" is replaced with the concept of bounded integer.

(2) the term "i" is replaced with the concept of an index.

(3) the term "i" is replaced with the concept of a step in an
induction (base case).

and so on. Let's forget about the mental dexterity required of one to replace a
word with a concept (quick, write down what the concept of a triangle
is...please be sure to reference Kant...now that you've written it down, are the
words you wrote the concept as a ding as sich?), there's no stopping the list of
concepts. And this is going to be done by a computer, no less, that doesn't even
understand natural language...assuming a machine understanding natural language
even makes sense.

Gerry

[ Reply to This | # ]

Protest...
Authored by: Anonymous on Friday, September 30 2005 @ 02:03 PM EDT
I don't understand this.

AFAICT this is considered to be a "published" patent application and because it's already "published" it's too late to file a protest with the USPTO. How were we supposed to learn about this patent and protest it BEFORE it was listed on the USPTO website?

[ Reply to This | # ]

International application for this patent is at WIPO.
Authored by: Anonymous on Friday, September 30 2005 @ 02:13 PM EDT

INTERNATIONAL APPLICATION PUBLISHED UNDER THE PATENT COOPERATION TREATY.... WIPO

What are the implications of this?

Brian S.

[ Reply to This | # ]

A patent with limited value
Authored by: Anonymous on Friday, September 30 2005 @ 03:56 PM EDT
Since the stated purpose of this application is to check proprietary code
against open source, it's only value may be to alert proprietary companies to
dangers they face from pilfered open source code.

[ Reply to This | # ]

Are there any patent attorneys around?
Authored by: Fredric on Friday, September 30 2005 @ 05:42 PM EDT
I am not a patent expert by far but I have read a couple of patents and to me this "patent application" does not look like anything I have seen. (We have on occasion seen Microsoft patents on Groklaw, read them and note the difference).

First, as some people pointed out, I can find no implementation. A patent should contain a description of how to implement the innovation, in this case it could be code or possibly free text. (Disclaimer: I could not persuade my browser to show the images in the patent so theoretically they could be the description but from the text I would expect a lot of boxes and arrows and no algorithms). AFAIK you can not patent an idea or concept (SCO would like that!) like a time-machine or the idea to create a web site to compare source code.

Second, the text does not look like something written by a patent attorney. I could of course be wrong but all patents I have read are much more precise and does not contain phrases like "it will be apparent to those skilled in the art" or similar (one can easily beleive that tha author does not consider himself "skilled in the art" ;-). Patent applications are usually much more difficult to understand because there is no reasoning or discussion and the reader is assumed to be not only skilled but an expert. This "patent application" feels more like a lecture than a patent application.

Anyway, this does not look real to me. I could not understand if the patent was granted. It would really surprise me if it was, but then, I am not a patent attorney so I could be wrong here and still get to keep my job.

Is there anybody around that can tell me if this really is a typical US SW patent application or does Mr. Anderer's elevator not really reach the top floor (if that translates).

---
/Fredric Fredricson
--------
[Funny sig temporarily removed for tests on Salisbury Plain]

[ Reply to This | # ]

Two months for prior art submission.
Authored by: rdc3 on Friday, September 30 2005 @ 08:04 PM EDT

As described in section 1134.01 of The Manual of Patent Examining Procedures, the USPTO allows a two month period for submission of prior art in the form of publications or patents, once an application is published.

Up to 10 publications or patents per submission may be filed. No accompanying explanations or highlighting is allowed. The fee is $180.

There is thus a very limited opportunity to file prior art to be considered by the patent examiner.

[ Reply to This | # ]

WOW! Google's Next up on SCO's sue list
Authored by: Anonymous on Friday, September 30 2005 @ 11:03 PM EDT
Directing your attention to Claim 3, which I paraphrase as "searching a set
of documents for a set of keywords and generating some figure of merit for the
goodness of the match."

It would appear that Google's, and presumably every other search engine that
generates goodness figures, is now an infringing application.

And you thought that SCO was only out after the Unix-like operating system
domain. I think that they had even bigger game in their sights!

Cheers!

[ Reply to This | # ]

Patent office shoots self in foot and gives away the store???
Authored by: Walter Dnes on Saturday, October 01 2005 @ 12:05 AM EDT
- Many patents (i.e. "process patents" and "business method
patents" and "software patents") cover "methods and
concepts".

- Anderer's application covers comparison on "methods and
concepts".

- A patent search involves comparing the text (source code) of patents to look
for the same "methods and concepts".

So the granting of this patent would ultimately mean that anyone, *INCLUDING
THE PATENT OFFICE*, doing a patent search would be infringing on this patent.
Another item to remember is that the first reaction by the Open Source community
to junk patents, is a search for prior art, looking for the same methods and
concepts. *SEARCHING FOR PRIOR ART AGAINST A PATENT GRANT WOULD BE AN
INFRINGEMENT OF THIS PATENT*!!!

If this patent is granted, the US patent system ...
1) is truly b0rk3n
2) will literally be 0wn3d... by Anderer

[ Reply to This | # ]

Mike Anderer, "MIT" Scientists, Spectral Analysis, and a Patent Application
Authored by: Anonymous on Saturday, October 01 2005 @ 10:02 AM EDT
"In this aspect, the input file may be patent claims or design specifications containing concepts that are compared against the concepts in the source code in the database. Thus, while the structure of the two corpuses is different (patent claims on the one hand and source code on the other) it is still possible according to the system of the present invention to determine whether they share concepts in common."

Declare two variables named "said" and "aforementioned", and this brilliant natural language processing system will conclude you violate every single patent in history.

[ Reply to This | # ]

What is IPX?
Authored by: Anonymous on Saturday, October 01 2005 @ 11:29 AM EDT
From the Anderer email:
"4) On the patent side for IPX, where foes that fit it. I am working with
the lawyers to get these moved from provisional to more complete in the next
week. I think it will spawn at least 3 patents. Ed and I are the inventors on
these."

ESR guessed that it referred to http://www.ipxonline.com/, "a company that
exists to help set up patent litigation". I haven't seen any link to this
company yet. And I'd hate to think we're going to assume that since the email
used the words patent, IPX and lawyers in the same sentence that we are
beholding to search engines for the bulk of our intelligence.

I'm not all-knowing either. But, wouldn't it be interesting to depose some of
the people involved and ask them if the "system for software code
comparison" was called IPX within their little clique?

[ Reply to This | # ]

  • What is IPX? - Authored by: Anonymous on Saturday, October 01 2005 @ 01:13 PM EDT
It's all in the name
Authored by: The Cornishman on Monday, October 03 2005 @ 08:38 AM EDT
> ...if our guess that it's the same man proves to be accurate

A literal translation of Anderer from the German: "Other". It's
clearly not the same man :)

---
(c) assigned to PJ

[ Reply to This | # ]

  • thank you - Authored by: Anonymous on Monday, October 03 2005 @ 09:49 AM EDT
Groklaw © Copyright 2003-2013 Pamela Jones.
All trademarks and copyrights on this page are owned by their respective owners.
Comments are owned by the individual posters.

PJ's articles are licensed under a Creative Commons License. ( Details )