W3CHINA.ORG讨论区--Trip report of FOWS 2009 and WWW 2009

I attended 4th Workshop on the Future of Web Search at Ibiza from April 17th to April 18th, 2009. The focus of FOWS 2009 is Semantic Search. It covers almost all topics related to semantic search including disambiguation and identification, query interface and result presentation, NLP & information extraction for semantic search, and retrieval and ranking.

As for the keynotes, External Mining of Search Query logs by Ziv Bar-Yossef, Google Haifa introduced how to use the limited results returned by general Web search engines to perform some common mining tasks such as query popularity and document popularity. The core technologies involved include sampling and probabilistic estimation. Part of the work has been published in WWW 2009 (see "Estimating the ImpressionRank of Web Pages" in the Graph Algorithms session).

The second keynote was presented by Hugo Zaragoza from Yahoo! Research Barcelona on "interacting with Semantically Annotated Collections". He introduced their work on making use of semantic annotations and NLP technologies to create richer interfaces for the search engine as well as improve the relevance of results. He also demonstrated their prototype system called Correlator (you can find it from Yahoo sandbox). You can also find more information from the invited talk "Correlator: things we did, things we should do, and things we don't know how to" of Semantic Search 2009 Workshop co-located with WWW 2009. You can learn more information about it from the invited talk of WWW 2009 by Ricardo Baeza-Yates, director of Yahoo research, Barcelona.

The final keynote speaker is Giovanni Tummarello from DERI, Galway. He is the unit leader of data incentive computing infrastructure and his talk is about "Scalable, Tolerant, Fair... ultimately useful: Web of Data processing for the benefit of Humans." He showed the vision on semantic search from DERI's perspective and introduced Sindice in details. From his talk, he focused on design choice and lessons learned during their implementation of Sindice. In particular, how to crawl most up to date semantic data, how to integrate lightweight reasoning (i.e. object consolidation according to inverse functional property), how to make use of cloud computing for distributed indexing and search, and how to leverage Yahoo BOSS to seamlessly combine semantic web search and web search.

Other interesting talks include
(1) Freebase: A socially managed identity database (by Jamie Taylor from Metaweb)
(2) Unique identifiers for the Web (by Zoltan Miklos from EPFL, Switzerland)
Note that this is funded by the EU FP7 project OKKAM and they have published a paper named "idMesh: Graph-Based Disambiguation of Linked Data" at WWW 2009.
3) Name ambiguity resolution and attribute extraction for the Web People Search task: overview of the WePS 2 evaluation campaign (by Julio Gonzalo from UNED Spain). Note that he organize a full-day workshop on Web people search at WWW 2009
4) Approximately Optimal Facet Selection (by Ronny Lempel from Yahoo research Hiafa).
5) Video shot retrieval : Who kills the vampire? (by Koen Deschacht from University of Leuven, Belgium)
6) Language-Model based Ranking in Entity-Relationship Graphs (by Shady Elbassuoni from Max-Planck-Institute for Informatics, Germany)
7) Data Web Search: What is to be done? (by Thanh from AIFB, Germany and me from Apex lab, Shanghai Jiao Tong University). Our presentation is similar to that made at the 2nd China Semantic Web Symposium. This time, we focused more on what to be done for semantic search.

Note that five work including "Semantically enhanced Information Retrieval: an ontology-based approach", "Investigating the Semantic Gap", "Concept Search: Enabling Semantics in Syntactic Search", "Relevance feedback for Semantic Search" and "Extracting structured data from text with applications" also made presentations in the Semantic Search workshop 2009.

After FOWS 2009, I also attended the 18th international World Wide Web conference from April 20 to 24 at Madrid. As one of the organizers, we held a very successful workshop on semantic search. I met several old friends there and some new friends since FOWS 2009. During the workshop, we decided to organizing an evaluation initiative for semantic search. We plan to make it more like INEX (community-based evaluation) rather than TREC due to the cost. I think it is quite important for pushing semantic search into practice.

As for the invited talks at WWW 2009, I enjoy the talk by Dr. Alfred Z. Spector from Google on "The Continuing Metamorphosis of the Web". He brought a novel perspective from Google on hybrid intelligence (artificial intelligence + human intelligence). In fact, it is quite similar to the idea of social semantic web.

During the main conference, I listened to search UI session, query processing session, caching and indices session, linked data session and ads and query expansion session. Most of them belong to the search track and one is from the semantic web track. My overall feeling is:
* It covers diverse topics on the Web. For example, Web privacy, Web engineering, mobile and rich media. It helps you to broaden your knowledge and experience.
* search and mining are two dominant topics in WWW. They occupy three tracks (search, data mining and social). Even in the semantic web track, one session is about mining for semantics.
* Linked data has caught much attention from both Semantic Web community and other communities. In the talk "Twenty Years: Looking Forward, Looking Back" by Tim Berners-Lee, he emphasized a lot on linked data along with some figures to show the fast development of linking open data.
* Monetization has been an emerging topic since WWW 2008. In particular, it is about advertising including auction, behavior targeting and sponsored search. You can regard it as a cross-field research area.

For more information in detail, you can refer to the homepage of FOWS and WWW. Wish you enjoy my trip report.


	W 3 C h i n a ( since 2003 ) 旗下站点苏ICP备05006046号《全国人大常委会关于维护互联网安全的决定》《计算机信息网络国际联网安全保护管理办法》	62.500ms