<img alt="" src="https://secure.leadforensics.com/98793.png" style="display:none;">

Scaling Intelligent Trademark Analysis

Scaling intelligent trademark analysis from prototype to production:

On Monday, I gave a presentation titled 'Scaling Intelligent Trademark Analysis from Prototype to Production: From MOSONG to Onomatics Quick Search' at the First International Workshop on Artificial Intelligence and IP Law (AIIP, see CfP and programme) arranged at the JURIX 2012 conference and supported by CREATe. Here is the abstract:

MOSONG is a high-fidelity model of legal decision-making in cases involving trademark similarity (likelihood of confusion) in EU law using type-2 fuzzy logic. The work done on MOSONG has also laid the foundation for the intelligent trademark analysis system developed by Onomatics, Inc., launched online in September 2012. There are however also many crucial differences between these two. For example, MOSONG still relies highly on the human element, that is, the correctness of its results is dependent on the domain knowledge of the user in addition to that embedded in the system, as evidenced by experimental testing on a large number of non-expert users. Using the Onomatics system, on the other hand, does not require any specific legal skills. This paper outlines this aspect and some of the other differences, which may be of broader interest as the steps necessary for transforming an academic proof-of-concept prototype into a marketable software product. The differences in the automated quantitative testing process for both systems are also highlighted.

The full papers won't be published until next autumn, but in the meantime you can take a look at my slides. Attendance could have been higher, so if you are an AI & law person looking for an application domain and could use some arguments for why you should choose specifically IP law, in anticipation of the next workshop (probably sometime during the summer), here are the arguments from my last slide once more:

IP law is an excellent legal domain for basic research in legal AI:
  • vague terms like similarity, originality, fair use, prior art in a central role, a challenge far too much ignored in AI & law so far
  • a field of law computer geeks love to hate (and know all too well)
  • based on international treaties: relevant, consistent and easily understandable around the world

And please remember that the workshop is not only for AI research on IP law but also for research in IP law doctrine concerning AI and its applications. (In fact, my own contribution next time around will probably belong to the latter category, for a change.) 


by Anna Ronkainen.