TextWise offers $1million for an American semantic hack

TextWise offers $1million for an American semantic hack

Summary: Erick Schonfeld at TechCrunch draws my attention to SemanticHacker. The site details an invitation from Rochester, NY-based TextWise to suggest compelling applications powered by their API, in return for a guaranteed payment of $100,000 and up to $900,000 in revenue from subsequent commercialisation of the winning idea.

SHARE:
TOPICS: Tech Industry
7

semantichacker-logo_200×29shkl.pngErick Schonfeld at TechCrunch draws my attention to SemanticHacker. The site details an invitation from Rochester, NY-based TextWise to suggest compelling applications powered by their API, in return for a guaranteed payment of $100,000 and up to $900,000 in revenue from subsequent commercialisation of the winning idea.

The challenge starts today, and runs until 18 June 2008. Entrants must be based in the United States, which rather unfortunately excludes the Semantic Web research powerhouses in Europe and Asia.

Quoting from the site;

"What will make you a winner in the SemanticHacker Innovators' Challenge?

  • Develop a software prototype, business plan or both that will have demonstrable commercial viability and the potential for significant financial impact on the application space to which it is applied.
  • Focus your submission on a vertical market. Areas such as finance, health and pharmaceuticals are just a few of the industries that might be a good place to start."

At the heart of TextWise's technological offer is an API they describe as "the world's first open API for Semantic Discovery." That strikes me as an assertion that probably needs to be ringed with caveats and footnotes if it's to stand up to closer scrutiny. This API enables developers to draw upon SemanticSignatures, defined as;

"a representation of ALL concepts covered in a block of text. Each block of text contains semantic dimensions ("concepts") with associated weights. The dimensions capture the strength of each concept in the text.

Semantic Signatures provide a weighted representation of the concepts contained in a piece of text. The weight of each concept represents the strength of that concept in the text. The Semantic Signatures for two pieces of text that both address the same subject will share many common concepts with high weights. Our technology can therefore recognize that these two pieces of text are related even though they share no common keywords."

It will be interesting to see what sort of entries this competition attracts, and the fact that TextWise have taken this course speaks volumes to the increasingly crowded semantic text analysis market. There's a lot of consolidation to come in this market segment, and the current players are working hard to draw attention to themselves. I, for one, would welcome some more effort devoted to explaining why they're different.

Topic: Tech Industry

Paul Miller

About Paul Miller

Paul Miller provides consultancy and analysis services at the interface between the worlds of Cloud Computing and the Semantic Web.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

7 comments
Log in or register to join the discussion
  • Semantic Hack

    Hi,

    Just pointing out that the entire semantic hack project
    may be simple theft, see www.tsert.com/white-papers/nlp.pdf

    Regards,
    Tsert.com
    Pierre Innocent
    pierreir
    • trouble reaching noted site tsert, but found way

      It's early on March 20, 2008. 2:30 AM PST

      The direct link to the Tsert site failed form me. I had success after Googling Tsert and trying what seemed an unlikely link:

      http://www.geocities.com/babelart/

      this page has only a text link, the same base URL as given in message, but which succeeded in passing me to the Tsert site and the PDF referred to.

      If there is theft, or some form of duplication of work, posting a color coded highlighted version of the text and/or PDF would help eliminate confusion. The PDF was a succinct summary of the fields of semantics, linguistics and mathematics, involved in Tsert's products, but no mention of solicitation for new ideas.

      Greg
      gregricemail@...
    • Semantic Hack - Patent References

      Mr. Innocent,

      I suggest you review our patents in this area:

      US Patent 6,751,621 - 06/15/2004

      Construction of trainable semantic vectors and clustering, classification, and searching using trainable semantic vectors

      http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsearch-bool.html&r=8&f=G&l=50&co1=AND&d=PTXT&s1=6,751,621&OS=6,751,621&RS=6,751,621

      US Patent 7,299,247 - 11/20/2007

      Construction of trainable semantic vectors and clustering, classification, and searching using trainable semantic vectors

      http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsearch-bool.html&r=1&f=G&l=50&co1=AND&d=PTXT&s1=6,751,621&OS=6,751,621&RS=6,751,621

      Regards,

      Wen Ruan
      Chief Science Officer
      TextWise
      wen_ruan
  • RE: TextWise offers $1million for an American semantic hack

    Give me a break! A company thinks they have a great tool, and they have to pay somebody $1M to come up with an idea to use it? This doesn't compute. Are you sure it's not a scam?
    w_c_mead
  • RE: TextWise offers $1million for an American semantic hack

    This has proven to be a very effective tactic in regard to garnering market share and improving a company's performance. I.e., encouraging outside input into your operations.

    One example I read about was a mining company up in Canada (don't remember the name of the company, sorry) that offered some huge prize to others, the mining company had all this geological research data and they offered that data for the contestants use, and the results were that they discovered several new highly productive mining sites.

    The value of the sites discovered were significantly higher than the cost of the prizes, and it opened up quite a few relationships of value with the contestants and their firms.
    Peopleunit
  • The rules are troubling and consistent with a scam

    The idea of the contest seems quite sensible. The people who invent any given tool might well not be the most talented users of it. Not to mention, the contest will likely increase public awareness of the tools and result in a large body of uncompensated experts evaluating a much wider field of potential applications than the in-house team could consider in the contest period. On the other hand, the provision inviting the judges to find no winner in the event the purposes of the contest are not met, seems to be a frank admission that the tools might actually be worthless.

    Curiously, no purpose is stated for holding the contest in the rules and no solid guidance given as to what a winning idea might look like, since the criteria stated in the rules can be adjusted or abandoned at any time without notice to the contestants ("Any of the above judging criteria is subject to change, modification, or elimination at the sole discretion of the Sponsor and/or the Panel and without any prior notification to the Entrant").

    Moreover, a close review of the contest rules raises numerous serious red flags in my opinion (and I should note I am a former attorney).

    The rules state that all entrants are granting the contest promoters a nonexclusive license to implement their ideas even if they are not selected as the winner ("Entrant shall grant TextWise an irrevocable, royalty-free, perpetual, non-exclusive worldwide right and license ... to use the entry for all uses"). Those licenses shall endure until the end of time ("perpetual"). Meanwhile, if 1,000 good ideas are received, the promoters are only required to pay for 1 of them. This provision seems grossly overreaching, but I have not participated in any similar contest and honestly don't know if this is a common contest rule or not.

    Apparently, any submission may be disqualified if the contestant refuses to assign ANY additional right demanded of them ("Entrants agree to any additional written authorization, assignment or other documentation requested by the Sponsor ..., and such written documentation will be required as a condition of selection"). I was expecting this term to contain the must complete any additional paperwork necessary to the purpose of these rules type language, but it actually expresses an entirely different and seemingly inappropriate burden on the contestant and does so only AFTER they have already disclosed their idea and granted the perpetual, nonexclusive license to use their idea to the sponsors.

    Even more bizarrely, the promoters are not obligated to pay the winner, because the promoters deny they have any obligation to notify the winner that they won ("Sponsor is not responsible for ... notification of the selected Entrants"). Hard to believe, but quite clear.

    Nor are the promoters obligated to pick any winner. ("The Panel may select one Entrant to receive the award ...." and "TextWise does not ... promise to select any Entrant") It might sound like nit-picking, but it is not. The "may/shall" distinction is a critical one among lawyers. A person must do what they are commanded they "shall" do. A person is under no obligation to do what they are commanded they "may" do.

    Additionally, the winner has only 16 business-hours ("two (2) business days") in which to obtain a legal consultation before being compelled to sign a series of complex legal documents, including a revenue-sharing agreement. Such a deadline is a very, very odd legal requirement. In 12 years of practice the only time anyone every gave me such a short deadline was when they were trying to strong-arm my client. While the revenue sharing agreement is available to review today, it is unreasonable to expect contestants to seek counsel before being declared the winner. Yet counsel is critical to evaluating whether one would want to sell all their rights for the lesser of 50% of first-year net revenues or $900k. Could the sponsors deflate net revenue by deliberately overinvesting in capital improvements to application infrastructure in the first 12 months? Are they serious when they say they have "no liability for the success or failure of its development, sales, marketing, licensing, distribution or other exploitation?" It sounds like they have no obligation to even try to be profitable in the first 12 months. This is not true though, because the law implies that term in the contract. In otherwords, the duty to use best efforts to maximize net revenue in the first 12 months exists without being mentioned and despite an apparent denial in the express terms of the revenue-sharing agreement. Clearly the winner needs a lawyer to evaluate the agreement.

    While nothing in the document prevents any contestant, except the winner if they elect revenue sharing, from setting up a competing business based on their submission, it is not difficult to imagine that the sponsors might be have a competitive edge and be capable of winning any marketplace contest, thus appropriating all the economic value of any submitted idea (other than the winning idea) for free.

    I thought the rules terms which assigned a license to the promoters might have been a poorly drafted reference to a license only to publish the entries, so as to demonstrate the potential virtues of the promoters' API, and not be meant to grant a license to use the technology describe in the entry, but that is not what is said. This is impression is strongly reinforced by the sentence following notice of the sponsors' right to publish submission materials, where it is stated, "Further, Sponsor and its affiliates shall have the right to otherwise use ... any Submission Materials."

    Further, the rules do not state when any such publication might occur, if at all (only that, "Submission Materials are non-confidential and may be made public"). Thus publication contrary to rules of patent eligibility (as I understand them - this is beyond my expertise though) could be made before a contestant learned that they had not won. Thus, anyone believing they have a patentable idea should probably patent it before submitting it. This considerably raises the cost of participation in the contest however, making it less palatable.

    Finally, the award committee may contain (or even be entirely comprised of) the same people who are paying for the award (the underwriting Venture Capitalists). One wonders what incentive they would have to pick the best idea as the winner, since picking a good, but economically weak, winner would allow them to avoid payment of $900,000 in contingency payments (though admittedly, it would open them up to otherwise avoidable competition).

    I want to stress that this contest may be completely legitimate, but the rules clearly entitle the promoters to some pretty unconscionable behavior. That does not mean they will act unconscionably though. I just wonder what the lawyer who wrote the rules was thinking. I cannot see how a competent drafter could have been ignorant of all the loopholes in the contest rules, nor of the bad smell they create in the minds of potential contestants.

    I would like to participate in this contest myself, but find myself much less inclined to do so having reviewed the rules.

    If we assume that honest men signify their honesty through the tender of fair contracts, then it would seem reasonable to assume the promoters of this contest are not honest men.
    gbambo
    • Response to gbambo

      Thank you for your feedback on the SemanticHacker Innovators' Challenge. We have forwarded your comments to our legal counsel for review.

      If any changes are made to the official rules, we will notify our users on the blog at http://blog.semantichacker.com.
      rpovio