Advances in Information Retrieval: 35th European Conference by Aleksandr Chuklin, Pavel Serdyukov, Maarten de Rijke

By Aleksandr Chuklin, Pavel Serdyukov, Maarten de Rijke (auth.), Pavel Serdyukov, Pavel Braslavski, Sergei O. Kuznetsov, Jaap Kamps, Stefan Rüger, Eugene Agichtein, Ilya Segalovich, Emine Yilmaz (eds.)

This publication constitutes the court cases of the thirty fifth ecu convention on IR examine, ECIR 2013, held in Moscow, Russia, in March 2013. The fifty five complete papers, 38 poster papers and 10 demonstrations offered during this quantity have been rigorously reviewed and chosen from 287 submissions. The papers are equipped within the following topical sections: person elements; multimedia and cross-media IR; info mining; IR idea and formal versions; IR method architectures; type; net; occasion detection; temporal IR, and microblog seek. additionally incorporated are four educational and a pair of workshop presentations.

Show description

Read or Download Advances in Information Retrieval: 35th European Conference on IR Research, ECIR 2013, Moscow, Russia, March 24-27, 2013. Proceedings PDF

Best european books

On Line Citizenship: Emerging Technologies for European Cities

On-line Citizenship discusses the evolutionary traits of towns when it comes to e-government within the current and destiny technological state of affairs. the focal point is at the transformation of the relationships among urban administrations and voters as a result of details and communique know-how. On a extensive viewpoint, the publication intends to provide possibilities and threats within the improvement of on-line citizenship within the framework of the eu Union.

Madness, love and tragedy in nineteenth and twentieth century Spain

How do Spanish writers of the nineteenth and twentieth century outline and symbolize insanity, a simple and debatable point of worldwide tradition, and the way do different conceptions of insanity intersect with love, faith, politics, and different literary topics in Spanish society? This multi-author publication analyzes the subject matter of insanity in formative masterpieces of Spanish literature of the nineteenth and twentieth century by using proper serious and theoretical ways.

Additional resources for Advances in Information Retrieval: 35th European Conference on IR Research, ECIR 2013, Moscow, Russia, March 24-27, 2013. Proceedings

Sample text

Using Intent Information to Model User Behavior in Diversified Search 7 The results, again in terms of perplexity gain, are shown in Fig. 3(b). Because we have much less data (121,431 sessions corresponding to 42,049 unique queries) our bootstrap confidence intervals are wide, wider than in Fig. 3(a). From the plot we see that only including layout information does not help, and that the best model in this situation is UBM-intents, which affirms our hypothesis. Gain per rank. The results so far report on perplexity gains over the complete SERP.

We can see that our improvements are statistically significant. 6 A. Chuklin, P. Serdyukov, and M. de Rijke Table 1. Average perplexity gain for the combined UBM-IA model Model Average Perplexity Gain Confidence Interval (Bootstrap) UBM-IA vs. 1 % combined layout intents only (a) Fresh results with special snippets combined layout intents only (b) Fresh results with ordinary snippets Fig. 3. Perplexity gains for layout and intent models compared to UBM Layout and intent in isolation. When we take a look at the modifications implemented on top of the UBM model, (4)–(6), we can see that they are actually a combination of two ideas: information about layout {Gk } and information about user intents I.

In: WSDM. : Improving personalized web search using result diversification. In: SIGIR. : Evaluating diversified search results using per-intent graded relevance. In: SIGIR. : Recency ranking by diversification of result set. In: CIKM, pp. 1949–1952. ACM (2011) 12 A. Chuklin, P. Serdyukov, and M. de Rijke A Appendix We describe an Expectation-Maximization algorithm for the UBM-IA click model. Algorithms for the other models considered in the paper can be derived in a similar manner. In order to simplify our calculations we use two sets of hidden variables: {Ek } (the user examined the k-th document) and {Ak } (the user was attracted by the k-th document).

Download PDF sample

Rated 4.85 of 5 – based on 43 votes