Tuesday, December 8, 2009

Microsoft Semantic Engine!

One of the highlights of the PDC09, an event focused on the technical strategy of the Microsoft developer platform, was the overview and demonstration of the Microsoft Semantic Engine. The Semantic Engine unifies search, structured querying, and analytics over structured and unstructured data.You can read some more details about PDC at the CTO's blog. It goes beyond existing components like Lucene by supporting both text and non-text, such as audio, video, and images. The key points are:

  • Micorsoft has been working and investing heavily on this technology for the last two years
  • The "Microsoft Semantic Engine" name is just a place holder
  • It is not a W3C SemanticWeb(tm) approach but one which melds the unique capabilities of unsupervised machine learning (hierarchical clustering), information retrieval models (higher-dimensional vector spaces), pluggable and trainable classifiers (SVMs, Naive Bayesian, Maximum Entropy, Decision Tree, etc.), and personalized filtering and ranking.
  • One of the goals is to make search semantically enhanced. Clustering the results based on Semantics is a key differentiator.
  • You can expect to see the Microsoft Semantic Engine in one of the upcoming SQL Server Betas.
This demo/presentation will explain the approach in more detail. My initial thoughts are:

  • This is a good move by Microsoft and was long overdue because there is a solid business case for integrating this technology in the enterprise.
  • I view it more as a flavor of text extraction/analytics technology - infact, Msft has said it very clearly that it is not using the semantic web technology approach
  • Having Microsoft in the Semantic technology space is a good thing for the Semantic Technology indutsry. This is a great validation from one of the most successful leaders in the software space
  • They are not the only one who is trying this approach. As far as I understand, the goal behind Inxight's acquisition by Business Objects (now SAP) was the same one. I am not aware exactly how that integration has worked out
  • I am sure Micosroft will come out with a clear message regarding the "Semantic Engine's" positioning in the enterprise in comparison to Fast Search Engine (Now part of Sharepoint division).
  • It will intersting to understand if Powerset (now Bing) technology, Microsoft's semantic search, was used in this effort
  • More details are needed to understand how it will work with disparate data sources
  • Microsoft has been underestimated for too long as far as their search strategy is concerned. I agree that Google has a big lead in terms of number of users on the web but I really think that they have made all the right moves, both for web as well as enterprises,at least  in last 2-3 years as far as search space is concerned. There are three great moves:
    • Acquistion of Fast search and integrating it with Sharepoint (MOSS)
    • Acquistion of Powerset and launch of Bing
    • Plan to introduce Semantic Engine and integrating it with SQL serve 
  • Probably,  services and product-based companies in the Semantic space, need to go back and revise their marketing message.
  • In the end, it just validates my initial thoughts that Semantic technology is a superset of Semantic Web Technology. Microsoft's approach or Semantic Web technology (W3C) approach are two different approaches to solve Semantic related problems in the enterprise. This is the reason, I named my blog as Semantic Technology blog.
Reblog this post [with Zemanta]

1 comment:

  1. I appreciate your post, thanks for sharing the post, i would like to hear more about this in future