![](bg6.png)
another to evaluate the SRQ (State Reformulated
Queries):
We asked 10 different users to submit 3 queries
(for doing different tasks), the system then detects
the task for each query. Next the users are asked if
the tasks were similar to their tasks or not. We then
got nearly 21 out of 30 positive responses (70%). To
evaluate the SRQ queries we asked the 10 users to
submit different queries and we applied each one to
the Google search engine at the different states of
the task which was proposed by our task model. We
reformulated these queries by adding the relevant
terms and then we reapplied them at the states using
the same search engine. We compared the first 20
retrieval results produced in the two cases (by
queries q and queries SRQ).
Results: we calculated the average number of
relevant pages by queries q and SRQ on the first 20
results (P@20). We noticed that the precision of the
relevant results using the initial query q is 0.17 and
0.59, respectively, by using SRQ queries which were
reformulated depending on the current task state and
user profile.
Conc urrent Substate
Before the Trip
Book the ticket
Hotel Reservation
preparation of the program
Concurrent Substate
After the Trip
news
pictures on the web
Concurrent Substate
During the Trip
find restaurant
Figure 3: Shows an example of a “travel task” that is
modeled by UML state diagram.
6 CONCLUSIONS
In this paper, we have proposed a hybrid method to
reformulate user queries depending on a dynamic
ontological user profile and user context for
producing State Reformulated Queries (SRQ). The
user context is considered as the actual state of the
task that he is undertaking when the information
retrieval process is performed. We have constructed
a general architecture that combines several models
for query expansion: the task model, the contextual
model, the user profile retrieval model and SRQ
model. We exploit both a semantic knowledge (ODP
Ontology) and a linguistic knowledge (WordNet) to
learn user’s task, and we exploit a UML states
diagram for this task to learn user current state. We
have also constructed a new general language model
for query expansion including the contextual factors
and user profile. We have illustrated on an
experimental study that the results obtained by SRQ
queries are more relevant than those obtained with
the initial user queries in the same task state. As a
future work, we plan to evaluate this method by
creating a test collection.
REFERENCES
Asfari, O., Doan, B. L., Bourda, Y. Sansonnet, J. P.:
Personalized access to information by query
reformulation based on the state of the current task and
user profile, In: The Third International Conference
on Advances in Semantic Processing, Malta, 2009.
Asfari, O.: Modèle de recherche contextuelle orientée
contenu pour un corpus de documents XML. In:
CORIA 2008: 377-384, France.
Bouchard, H., Nie, J. Y.: Modèles de langue appliqués à la
recherche d’information contextuelle, In: Conf. en
Recherche d’Information et Applications (CORIA),
Lyon, 2006.
Bhogal, J., Macfarlane, A., Smith, P.: A review of
ontology based query expansion, Information
Processing and Management. In: an International
Journal, v.43 n.4, p.866-886, July, 2007.
Freund, L., E. G. Toms, and Charles, L. A. Clarke:
Modeling task-genre relationships for IR in the
workplace, In: SIGIR 2005, at Salvador, Brazil.
Koutrika, G., Ioannidis, Y. E., Personalization of Queries
in Database Systems, In: Proceedings of 20th
International Conference on Data Engineering, 2004.
Lv, Y., Zhai C.: Adaptive Relevance Feedback in
Information Retrieval, IN: CIKM, 2009, Hong Kong.
Luxenburger, J., Elbassuon, S., Weikum, G.: Task-aware
search personalization, In: Proceedings of the 31st
annual international ACM SIGIR, Singapore, 2008.
Micarelli, A., Gasparetti, F., Sciarrone, F., and Gauch, S.:
Personalized Search on the World Wide Web. P.
Brusilovsky, A. Kobsa, and W. Nejdl (Eds.), In: The
Adaptive Web, LNCS 4321, Berlin, 2007.
Mylonas, Ph., Vallet, D., Castells, P., Fernandez, M., and
Avrithis, Y., Personalized information retrieval based
on context and ontological knowledge. In: Knowledge
Engineering Review, Cambridge University, 2008.
Navigli, R., Velardi, P.: An Analysis of Ontology-based
Query Expansion Strategies, In: Workshop on
Adaptive Text Extraction and Mining at the 14th
Conference on Machine Learning, Croatia 2003.
Sieg, A., Mobasher, B., and Burke, R.: Representing
context in web search with ontological user profiles.
In: Proceedings of the 6th International Conference on
Modeling and Using Context, Denmark, August 2007.
Terai, H., Saito, H., Takaku, M., Egusa, Y., Miwa, M., and
Kando, N.: Differences between informational and
transactional tasks in information seeking on the web.
In: Proceedings of the Second Symposium IIiX, 2008.
W. White, R., Kelly, D.: A Study on the Effects of
Personalization and Task Information on Implicit
Feedback Performance, In: CIKM’06, 2006, USA.
A CONTEXT-BASED MODEL FOR WEB QUERY REFORMULATION
293