Search the KHIT Blog

Tuesday, August 29, 2017

NLP, meet NPE: the curious case of Robert Budzinski


Robert Lucius Budzinski, of 1106 Edgewood Drive, Richardson, TX 75081, patent infringement plaintiff asserting to be the "sole inventor" of the Natural Language Processing (NLP) technology alleged wrongfully used by a breadth of major tech corporations. He is reported to be suing a number of them, e.g., Nuance, Google, Apple, Microsoft, and Amazon.

"NLP, meet NPE"

"NPE" is intellectual property legalese for "Non-Practicing Entity."

Otherwise known as a "patent troll." Mr. Budzinski is the owner/"President"/"Director"/"Registered Agent" (and sole "employee") of "Word to Info, Inc.," a private corporation (operating out of his residence) that apparently produces no discernible products or services (a "NPE"). A "company" with no internet presence of any sort -- if you exclude Mr. Budzinski's LinkedIn page:


That's it in its entirety. No employment or education history, no accomplishments or interests, no connections, no nothing. Zip, zilch, nada, nyet.
I did find Plaintiff's Counsel info in one of the lawsuit filings:
Steven R. Daniels
Texas State Bar No. 24025318
FARNEY DANIELS PC
800 South Austin Ave., Suite 200
Georgetown, Texas 78626
Email: sdaniels@farneydaniels.com
Telephone: (512) 582-2828
Fax: (512) 582-2829


James R. Gourley
Texas Bar No. 24050679
CARSTENS & CAHOON, LLP
13760 Noel Road, Suite 900
Dallas, Texas 75240
Email: james@cclaw.com
Telephone: (972) 367-2001
Fax: (972) 367-2002
Attorneys for Plaintiff
Word to Info, Inc.
I routinely search Google news and other internet sources for anything new on "natural language processing" (NLP), looking in particular for stuff relating to the health care space. Recall my prior posts on the topic, here, here, and here.

I ran across this at AppleInsider.com:
Apple's Siri latest target in string of natural language patent lawsuits
By Mikey Campbell


One-man company Word to Info on Friday expanded a string of patent lawsuits over natural language processing technology — active cases involve Amazon, Google, Microsoft and Nuance — to include Apple, taking specific aim at the tech titan's Siri virtual assistant.

Filed in the patent holder-friendly Eastern Texas District Court, Word to Info's suit alleges infringement of seven patents detailing methods of natural language processing. The company asserted the same series of patents against a number of tech industry giants marketing their own voice recognition and virtual assistant solutions.

Specifically, Word to Info is leveraging U.S. Patent Nos. 5,715,468, 6,138,087, 6,609,091, 7,349,840, 7,873,509, 8,326,603 and 8,688,436 in its case against Apple. The IP string covering methods of interpreting natural language input dates back to 1998, when the U.S. Patent and Trademark Office granted the '468 property to inventor and Word to Info director Robert Budzinski...
I did a cursory review of all seven patent filings listed in the article. They total about 2,100 pdf printable pages of dense detail -- most of the content repetitiously redundant.
UPDATE: the most recent (08/18/17) TX court filing -- against Apple -- by this plaintiff here (49 pg pdf). The suit alleges seven "Claims for Action," each of them mapping to the Budzinski patents set forth and linked above.
The "abstracts" from all of the Budzinksi patents, from the oldest to the most recent:
Abstract
A memory system for storing and retrieving experience and knowledge with natural language. The primary components of this memory system include syntactic processes, function word processes, morphology processes, ellipsis processes, concrete and abstract noun word sense number processes, verb word sense number processes, adjective word sense number processes, purpose identification processes, plausibility and expectedness processes, communication processes, context storage processes, and text generation processes. The syntactic processes include word isolation, dictionary look up, and parsing. The function word processes select and evaluate functions associated with function words which are certain: adjectives, nouns, verbs, adverbs, pronouns, prepositions, conjunctions, and interjections. Morphology processes replace morphological words with phrases or clauses composed of function words arid state representation words. Certain nouns, verbs, and adjectives are state representation words. Selecting the word sense number of a state representation word selects the state representation of such a word. Experience and knowledge are stored as clause implying word sense numbers organized into paths in a directed graph.

Abstract
A memory system for storing and retrieving experience and knowledge with natural language through methods and apparatus is disclosed. The primary components of this memory system include syntactic processes, function word processes, ellipsis processes, morphology processes, meaning word sense number processes, purpose identification processes, plausibility and expectedness processes, communication processes, context storage processes, and text generation processes. The function word processes select and evaluate functions associated with function words which are certain words in each part of speech. Ellipsis processes replace unstated words. Morphology processes replace morphological words with phrases or clauses composed of function words and word sense numbers. A word sense number is an address to the meaning of a word. Certain nouns, verbs, and adjectives are meaning words. Word sense numbers are selected to be consistent with the context of the clause containing the words, the context, and stored experience and knowledge. Experience and knowledge are stored as nodes with associated clause implying word sense numbers organized into paths in a directed graph. Nodes in the directed graph have access conditions which determine if a node is accessible on a path. A path has an associated purpose relation which is any concept that labels the path. Purpose identification processes select: purpose relations, experience and knowledge, processes for setting a verb's result states or a state value, object classification paths, and activity selection paths. The communication processes coordinate incoming and outgoing natural language text. Text generation processes generate natural language text from word sense numbers.

Abstract
A memory system for storing and retrieving experience and knowledge with natural language through methods and apparatus is disclosed. The primary components of this memory system include syntactic processes, function word processes, ellipsis processes, morphology processes, meaning word sense number processes, purpose identification processes, plausibility and expectedness processes, communication processes, context storage processes, and text generation processes. The function word processes select and evaluate functions associated with function words which are certain words in each part of speech. Ellipsis processes replace unstated words. Morphology processes replace morphological words with phrases or clauses composed of function words and word sense numbers. A word sense number is an address to the meaning of a word. Certain nouns, verbs, and adjectives are meaning words. Word sense numbers are selected to be consistent with the context of the clause containing the words, the context, and stored experience and knowledge. Experience and knowledge are stored as nodes with associated clause implying word sense numbers organized into paths in a directed graph. Nodes in the directed graph have access conditions which determine if a node is accessible on a path. A path has an associated purpose relation which is any concept that labels the path. Purpose identification processes select: purpose relations, experience and knowledge, processes for setting a verb's result states or a state value, object classification paths, and activity selection paths. The communication processes coordinate incoming and outgoing natural language text. Text generation processes generate natural language text from word sense numbers.

Abstract
A memory system for storing and retrieving experience and knowledge with natural language through methods and apparatus is disclosed. The primary components of this memory system include syntactic processes, function word processes, ellipsis processes, morphology processes, meaning word sense number processes, purpose identification processes, plausibility and expectedness processes, communication processes, context storage processes, and text generation processes. The function word processes select and evaluate functions associated with function words which are certain words in each part of speech. Ellipsis processes replace unstated words. Morphology processes replace morphological words with phrases or clauses composed of function words and word sense numbers. A word sense number is an address to the meaning of a word. Certain nouns, verbs, and adjectives are meaning words. Word sense numbers are selected to be consistent with the context of the clause containing the words, the context, and stored experience and knowledge. Experience and knowledge are stored as nodes with associated clause implying word sense numbers organized into paths in a directed graph. Nodes in the directed graph have access conditions which determine if a node is accessible on a path. A path has an associated purpose relation which is any concept that labels the path. Purpose identification processes select: purpose relations, experience and knowledge, processes for setting a verb's result states or a state value, object classification paths, and activity selection paths. The communication processes coordinate incoming and outgoing natural language text. Text generation processes generate natural language text from word sense numbers.

Abstract
A memory system for storing and retrieving experience and knowledge with natural language through methods and apparatus is disclosed. The primary components of this memory system include syntactic processes, function word processes, ellipsis processes, morphology processes, meaning word sense number processes, purpose identification processes, plausibility and expectedness processes, communication processes, context storage processes, and text generation processes. The function word processes select and evaluate functions associated with function words which are certain words in each part of speech. Ellipsis processes replace unstated words. Morphology processes replace morphological words with phrases or clauses composed of function words and word sense numbers. A word sense number is an address to the meaning of a word. Certain nouns, verbs, and adjectives are meaning words. Word sense numbers are selected to be consistent with the context of the clause containing the words, the context, and stored experience and knowledge. Experience and knowledge are stored as nodes with associated clause implying word sense numbers organized into paths in a directed graph. Nodes in the directed graph have access conditions which determine if a node is accessible on a path. A path has an associated purpose relation which is any concept that labels the path. Purpose identification processes select: purpose relations, experience and knowledge, processes for setting a verb's result states or a state value, object classification paths, and activity selection paths. The communication processes coordinate incoming and outgoing natural language text. Text generation processes generate natural language text from word sense numbers.

Abstract
A memory system for storing and retrieving experience and knowledge with natural language through methods and apparatus is disclosed. The primary components of this memory system include syntactic processes, function word processes, ellipsis processes, morphology processes, meaning word sense number processes, purpose identification processes, plausibility and expectedness processes, communication processes, context storage processes, and text generation processes. The function word processes select and evaluate functions associated with function words which are certain words in each part of speech. Ellipsis processes replace unstated words. Morphology processes replace morphological words with phrases or clauses composed of function words and word sense numbers. A word sense number is an address to the meaning of a word. Certain nouns, verbs, and adjectives are meaning words. Word sense numbers are selected to be consistent with the context of the clause containing the words, the context, and stored experience and knowledge. Experience and knowledge are stored as nodes with associated clause implying word sense numbers organized into paths in a directed graph. Nodes in the directed graph have access conditions which determine if a node is accessible on a path. A path has an associated purpose relation which is any concept that labels the path. Purpose identification processes select: purpose relations, experience and knowledge, processes for setting a verb's result states or a state value, object classification paths, and activity selection paths. The communication processes coordinate incoming and outgoing natural language text. Text generation processes generate natural language text from word sense numbers.

Abstract
A memory system for storing and retrieving experience and knowledge with natural language through methods and apparatus is disclosed. The primary components of this memory system include syntactic processes, function word processes, ellipsis processes, morphology processes, meaning word sense number processes, purpose identification processes, plausibility and expectedness processes, communication processes, context storage processes, and text generation processes. The function word processes select and evaluate functions associated with function words. A word sense number is an address to the meaning of a word. Word sense numbers are selected to be consistent with the context and stored experience and knowledge. Experience and knowledge are stored as nodes with access conditions and with associated clause implying word sense numbers organized into paths in a directed graph. A path has an associated purpose relation which is any concept that labels the path. Text generation processes generate natural language text from word sense numbers.
Yeah, gave me MEGO as well. Beneath each abstract are dozens of equally dense enumerated "claims" ostensibly setting forth the putative (and obtuse) "operational details" of the "invention." The old jibe "if you can't dazzle 'em with brilliance, baffle 'em with bullshit" comes to mind.
BTW: I just asked Siri on my iPhone 6s (multiple times) "Is Robert L. Budzinski a patent troll?" "She" couldn't accurately process the query. All I got was repeated irrelevant gibberish.
Also of note: I never once encountered in these patent documents the phrase "computational linguistics" (or the word linguistics, for that matter).


Moreover, checking the body text and index of this authoritative reference book I have studied and heretofore cited, no mention of a "Budzinski." You would think that a guy purporting to be the "sole inventor" of a huge chunk of applied AI/NLP (worth potential billions) would have an academic and professional rap sheet a mile long, one that would show up in a boatload in scholarly citations.

Maybe I'm missing something. Not for lack of extensive searching.

UPDATE

You keep digging, using various name/word/phrase combinations and permutations, eventually you find something.


1981. Looks like he worked for T.I. Obviously some kind of microelectronics guy (engineer?). Link to the above paper here. I'd eventually run across an IEEE link citing him as as a co-author on three papers (including the above) around that time (the other two are firewalled in pdf at IEEE).

Still have yet to unearth a C.V. Still smells like "NPE," notwithstanding that his cred has to be up just a notch in my eyes, although technical cred is a separate issue from patent trolling.

There's a ton of reportage on patent trolling, (which has been recently asserted to "cost the economy" $80 billion a year), e.g.,
'THE ULTIMATE PATENT TROLL'
In the words of technology reporters.
MORGAN BASKIN & JACK DENTON


Founded in 2000, Intellectual Ventures "has earned a special brand of hatred in the business world as the ultimate patent troll. It doesn't delay your flight like United, buffer your movie stream like Comcast, or shellac your shrimp with oil like BP. Rather, it hoards ideas." It "goes around to companies and says: 'Hey, you want to protect yourself from lawsuits? We own tons of patents. Make a deal with us. Our patents will not only cover everything you're doing in your business, no one will dare sue you." "It then wields this intellectual-property portfolio—the world's largest—like a weapon. Companies can either pay up or face a lawsuit.”…
__

Back to Budzinski...

"PLAINTIFF DEMANDS A JURY TRIAL"

But of course he would.

In 1991 I took a gig as the Technical Editor (pdf)  for "CSI" ( Computational Systems, Inc., of West Knoxville TN). It was a small recent startup founded by two PhD industrial-electrical engineers who were alumni of a larger competitor (TEC). We designed, built, and marketed digital Fast Fourier Transform (FFT) industrial analyzers. We were staffed up with hardhat assembly and C language programmers writing tight code straight to the chip clocks.

A great, fun job. Heady stuff. Kind of an Appalachian "Silicon Valley" loosey-goosey culture producing then-state of the art portable industrial digital tech (we were just across the pike east of Oak Ridge, where I'd spent the prior five and a half years in a radiation lab writing code). My department was an "artsy" Mac shop, basically an in-house ad agency doing magazine-quality 4-color digital pre-press.

Well, our competitor sued us for patent infringement, based on asserted purloined IP (essentially a de facto "non-compete" "leave your brain at the door" case). We were killing them in sales (owing more to our assiduous customer service and support than to our technologies per se). They retaliated.

And they won. Jury trial. Crippling monetary judgment. Only one juror had beyond a high school education. [Technology For Energy Corporation v. Computational Systems, Inc., et al., (E.D. Tenn., Fed. Cir.) (Patent infringement litigation)]

I survived the ensuing layoffs.

Wasn't a patent troll thing, but I will never forget the turmoil. In July of 1992 I left to move to Las Vegas in the wake of my wife's transfer and promotion to manage QA at the Nevada Test Site nuke cleanup project.

I then took an analyst/LAN manager job with the Nevada Medicare QIO (HealthInsight, then known as Nevada Peer Review). I recall during my first stint (of three) with them a discussion one day concerning an ortho doc who did more than a quarter of all hip jobs in the state. Owing to the relatively high efficacy of his px, he sought to patent his surgical "method."

Lordy. The blowback was pretty severe. I don't think he ever went through with it.

BTW, a good, quick read on intellectual property fundamentals and issues:

"We all create intellectual property. We all use intellectual property. Intellectual property is the most pervasive yet least understood way we regulate expression. Despite its importance to so many aspects of the global economy and daily life, intellectual property policy remains a confusing and arcane subject. This engaging book clarifies both the basic terms and the major conflicts surrounding these fascinating areas of law, offering a layman's introduction to copyright, patents, trademarks, and other forms of knowledge falling under the purview of intellectual property rights..."
I finished this book in short order. Highly recommended survey look into the breadth of sub-topics of IP: patents, copyrights, trademarks, and trade secrets, as well as relatively minor tangentially overlapping areas such as fashion design, internet domain names, and celebrity exclusive use of "likeness." Good discussion of the evolved histories and global differences in IP scope and enforcement.
Recent item: General Mills just (rightfully) lost an IP lawsuit wherein they tried to obtain exclusive trademark rights to the color "yellow," on the grounds that market identification of Cheerios was/is inextricably bound up with it in the product's packaging. Consumers might otherwise be "confused."
The jokes just write themselves.
Jokes aside, we might ask, what is the proper scope of intellectual property ownership enforcement in the law?
If there were only one man in the world, he would have a lot of problems, but none of them would be legal ones. Add a second inhabitant, and we have the possibility of conflict. Both of us try to pick the same apple from the same branch. I track the deer I wounded only to find that you have killed it, butchered it, and are in the process of cooking and eating it.

The obvious solution is violence. It is not a very good solution; if we employ it, our little world may shrink back down to one person, or perhaps none. A better solution, one that all known human societies have found, is a system of legal rules explicit or implicit, some reasonably peaceful way of determining, when desires conflict, who gets to do what and what happens if he doesn’t…


Friedman, David D.. Law's Order: What Economics Has to Do with Law and Why It Matters (p. 3). Princeton University Press - A. Kindle Edition. 
I have not the slightest doubt that, were I to query Budzinski's patent attorneys, I'd be met with harrumphingly indignant, derisive (albeit utterly self-serving) paternalistic pushback lecturing down to me about the Altruistic Nobility of their tireless work in defense of "the little guy."


A planet populated by more than 7 billion contending people is a complicated place. Our technologies are complicated, ever-moreso. Necessarily, our laws are increasingly complicated. And, among the most complex among them are the laws going to IP. Nonetheless, to the extent that IP filings are woefully obtuse, dense documents riddled with verbose, redundant, vague, ambiguous, often internally inconsistent jargon-language, they seem to principally only serve the lawyers whom they mostly enrich.**
** A similar observation can be made in general about legislation and the regulations they authorize. Overcomplexity and lack of clarity in law inevitably beget legal challenges, where the courts have to try to clean up the messes.
For one thing, perhaps "Loser Pays" IP litigation reform might serve to mitigate the excesses in this area. (Ohhh... the poignant wailing that that notion produces in tortland!) I also have to wonder how IP litigation firms' expenses are covered prior to judgments or settlements? Is this stuff typically done on "spec?" And, are there the equivalent of "VC/investment funds" bankrolling these actions? I seriously doubt that our NLP boy Rob is paying out of pocket by the billable hour.

 THE NPE AT SILICON VALLEY HBO


LOL.

CODA

Pretty interesting read at The New Yorker:
Who Owns the Internet?
What Big Tech’s monopoly powers mean for our culture.
By Elizabeth Kolbert
...Thirty years ago, almost no one used the Internet for anything. Today, just about everybody uses it for everything. Even as the Web has grown, however, it has narrowed. Google now controls nearly ninety per cent of search advertising, Facebook almost eighty per cent of mobile social traffic, and Amazon about seventy-five per cent of e-book sales...
And Rob Budzinski wants him a piece of all that. "Rentier?"

Relatedly,
Google Just Proved That Monopolies Imperil Democracy, Not Just The Economy
Barry Lynn and his team of anti-monopoly researchers were fired by a think tank after criticizing the search giant.


WASHINGTON ― For the past decade, former business journalist Barry Lynn has used his perch at the New America Foundation to warn politicians and the public that a new era of corporate monopolies threatened not only American workers, but also democracy itself.

Lynn was just proven right: New America has fired him as head of its Open Markets program along with his team of about 10 researchers and journalists, after they called for an antitrust investigation of the think tank’s largest longtime donor, Google…
Naked Capitalism is also on the story: "New America Foundation Head Anne-Marie Slaughter Botches Laundering Google’s Money, Fires Anti-Trust Team at Eric Schmidt’s Behest."

OH, AND, ANOTHER THING,

apropos, read this in my AAAS Science Magazine:
Fostering reproducibility in industry-academia research
Science  25 Aug 2017:
Vol. 357, Issue 6353, pp. 759-761
DOI: 10.1126/science.aan4906


Many companies have proprietary resources and/or data that are indispensable for research, and academics provide the creative fuel for much early-stage research that leads to industrial innovation. It is essential to the health of the research enterprise that collaborations between industrial and university researchers flourish. This system of collaboration is under strain. Financial motivations driving product development have led to concerns that industry-sponsored research comes at the expense of transparency (1). Yet many industry researchers distrust quality control in academia (2) and question whether academics value reproducibility as much as rapid publication. Cultural differences between industry and academia can create or increase difficulties in reproducing research findings. We discuss key aspects of this problem that industry-academia collaborations must address and for which other stakeholders, from funding agencies to journals, can provide leadership and support…

Barriers to Sharing
Efforts to promote reproducible research have varied. One widely supported strategy is to increase the availability of data produced in studies, along with computer code written to clean and analyze data. Publishers and funders have instituted policies mandating data deposition or data management plans; however, success has not been uniform.

There are disincentives to open sharing of information. For academic research, rewards come from public presentations and publications that lead to recognition within the community, grants, and tenure. The emphasis on publications to reap academic rewards means that academic researchers can be reluctant to release information or even to fully describe their work. In industry, publishing is typically not a high priority; the goals are the provision of a product (whether goods or services) that will outstrip competitors and provide monetary rewards to investors. The need to obtain patents or maintain trade secrets to protect intellectual property (IP) can provide a strong financial incentive to not disclose or share information. Corporations see relatively little advantage to releasing data for research purposes, so any nonzero risk of consequences (even if only hypothetical) can be sufficient to shut down such efforts…

The Creative Commons license CC0 (which waives all rights of data authors) is attractive as it does not require data sets to have a “provenance” trail and can thus ease automated mining of data. However, lack of provenance tracking in CC0 creates challenges for data evaluation, interpretation of analyses, and accreditation of data generators, thus making CC-BY (in which author attribution is required) attractive. Both continue to be discussed as aspirational goals.

Irreproducible research wastes time, money, and resources. Academic researchers, universities, and other institutions, industry, funding agencies, and editors all have a role to play in raising research standards and creating an environment of trust between communities.
Hmmm... For one thing, I refer you to this book I cited a while back (scroll down):

____________

More to come...

No comments:

Post a Comment