A. Pretschner. "Ontology Based Personalized Search". Master 's thesis, Department of Electrical Engineering and Computer Science - University of Kansas, 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....a viable approach to overcome the problem of semantic heterogeneity. Interoperability between agents can be achieved reconciliating the different world views through the compromise to a common ontology [19] Also, information agents can use ontologies to represent explicitly isolated user profiles [17, 1], which is the view adopted in this work concerning the profiles generated by the PersonalSearcher agent. 2.2 PersonalSearcher The PersonalSearcher [5] agent is dedicated to the task of retrieving information from the web. It uses Case Base Reasoning to build a profile of the thematic ....

A. Pretschner. Ontology Based Personalized Search. MSc. Thesis. Kansas Univ., USA, 1999.

....The ontologies that we consider although simple, they fit to the content based organizational structure of Web catalogs and portals, keyword hierarchies and personal bookmarks. Besides most of the ontologies that are used for indexing and retrieving objects are term hierarchies ( 3] 4] [5]) Concerning the functionality offered by our mediators, the objective that governs the selection of the sources to be queried (and the formulation of the queries to be sent to each source) is to minimize the semantic difference between the received query and the query finally answered by the ....

Alexander Pretschner. "Ontology Based Personalized Search". Master's thesis, Department of Electrical Engineering and Computer Science - University of Kansas, 1999.

....of words that is different from the user s input. Ontology Based Personalized Search Because each keyword has a different mearning, depending on the context, search engines often return results that may be completely irrelevant to the user s needs. The Ontology Based Personalized Search [14] improves upon com merical search engines by including the user s profile in the matching process. The user s profile can be learned through explicit feedback about the quality of websites or observing the user s access patterns when surfing the web. The user s profile is also another feature ....

A. Pretschner. Ontology based personalized search. Master's thesis, University of Kansas, 1998.

....of our technique. We outline the implementation of our algorithm in the Persona personalized web search system. 1 Overview Search engines index large numbers of documents and let users query desired docu ments. However, most search engines are not tailored to meet individual user pref erences. [6] noted that almost half of the documents returned by search engines are deemed irrelevant by their users. There are several aspects to the problem. First is the problem of synonyms and homonyms. Synonyms are two words that are spelt differently but have the same meaning. Homonyms are words that ....

....The idea is to capture in the user profile sets of words that may have different meaning when coupled together (e.g bar tender) The paper also offers suggestions on which sets of words should be given more emphasis, such those within the bold tag and italic tag in an HTML page. Another work by [6] describes personalized search based on a taxonomy. It uses an existing taxonomy from Magellan, which classifies documents into approximately 4400 nodes. Profiles are stored as concept hierarchies, as in the SmartPush system. Each node is associated with a set of documents. Each document is ....

[Article contains additional citation context not shown here]

PRETSCHNER, A. AND GAUCH, S., Ontology Based Personalized Search, in Pro- ceedings of the Eleventh IEEE International Conference on Tools with Artificial Intelligence, 1999.

....of our technique. We outline the implementation of our algorithm in the Persona personalized web search system. 1 Overview Search engines index large numbers of documents and let users query desired documents. However, most search engines are not tailored to meet individual user preferences. [6] noted that almost half of the documents returned by search engines are deemed irrelevant by their users. There are several aspects to the problem. First is the problem of synonyms and homonyms. Synonyms are two words that are spelt di erently but have the same meaning. Homonyms are words that are ....

....The idea is to capture in the user pro le sets of words that may have di erent meaning when coupled together (e.g bar tender) The paper also o ers suggestions on which sets of words should be given more emphasis, such those within the bold tag and italic tag in an HTML page. Another work by [6] describes personalized search based on a taxonomy. It uses an existing taxonomy from Magellan, which classi es documents into approximately 4400 nodes. Pro les are stored as concept hierarchies, as in the SmartPush system. Each node is associated with a set of documents. Each document is ....

[Article contains additional citation context not shown here]

Pretschner, A. and Gauch, S., Ontology Based Personalized Search, in Proceedings of the Eleventh IEEE International Conference on Tools with Arti cial Intelligence, 1999.

....well with the content based organizational structure of web catalogs and portals (e.g. Yahoo , Open Directory 2 ) keyword hierarchies (e.g. ACM s thesaurus) and personal bookmarks. Besides most of the ontologies that are used for indexing and retrieving objects are term hierarchies ( 10] 14] [17]) Since many ontologies (e.g. those employed by web catalogs) usually contain very large numbers of terms, the articulation of ontologies has many advantages comparing to ontology merging. Clearly, merging the ontologies of all underlying sources would introduce storage and performance overheads. ....

A. Pretschner. "Ontology Based Personalized Search". Master 's thesis, Department of Electrical Engineering and Computer Science - University of Kansas, 1999.

....(Noy Hafner 1997) Clark 1999) Chandrasekaran, Josephson, Benjamins 1999) There is little agreement on a common definition of an ontology, most works cite (Gruber 1993) as the common denominator of all ontology definitions. Ontologies are used in user profiles, for information retrieval (Pretschner Gauch 1999), for databases (J.Bayardo et al. 1997) and agent communication (Huhns Singh 1997) As proposed in (Welty 2000) talking about ontologies only makes sense regarding a certain application context. In this paper we look at ontologies for categorizing documents by their contents. The YAHOO ....

Pretschner, A., and Gauch, S. 1999. Ontology based personalized search. In Proc. 11th IEEE Intl. Conf. on Tools with Artificial Intelligence, pp. 391--398.

....their Web sur ng behavior to be monitored and that information to be shared with a search engine, which will be a social barrier for many users. As it is not directly related to dimensionality reduction, this approach will not be further discussed in this paper, but a discussion can be found in (Pretschner and Gauch 1999). 2 The Curse of Dimensionality and Feature Reduction The natural features of text documents are words or phrases, and a document collection can contain millions of di erent features. Even after applying standard feature reduction techniques, the number of features remains large: in our ....

....a browsing structure automatically by categorizing documents into concepts in a pre de ned ontology, or subject hierarchy. In one application, all the pages for a given Web site are categorized into concepts in a standard ontology to produce a site map for that site (Zhu, Gauch, Gerhard, Kral and Pretschner 1999). A site map represents the information space of a site by either relating a group of Web pages to a particular subject or depicting the links speci ed in the Web pages. To generate a subject based site map, one must describe a subject space, which typically has a hierarchical structure, and then ....

Pretschner, A. and S. Gauch (1999). Ontology based personalized search. In Proc of the 11th IEEE International Conference on Tools with Articial Intelligence (ICTAI '99), Chicago, IL, pp. 391{ 398.

....by a search engine, every word contained in them is looked up in the pro le. If it exists in the user pro le, its weight in the retrieved document is added to the URLs score. This yields a new personalized ranking. Architecture: Pro les are stored on the client machine. The system described in [44] aims at improving search results by re ranking and ltering them. Pro le representation, Information Sources: Pro les are created as a function of the web sur ng history of an individual user. Surfed pages are characterized (i.e. their content or descriptive categories are determined) w.r.t. to ....

....what how ad. st. rating model coll. ind. WebWatcher browsing ass. keywords repr. interests goals, links annotated with prof. reinforcem. learning st. coll. Yahoo pers.access, portal keywords st. ind. 27] search local le system, stored as keywords st. based on freq. ind. [44] search sur ng behavior, ontology of 4,300 nodes ad. structured cosine sim. ind. 57] expertise location JAVA source codes st. coll. Table 1: Systems with personalization services 4 Discussion Personalization is a very active and broad area of research with many applications. The main ....

A. Pretschner and S. Gauch. Ontology Based Personalized Search. In Proc. 11th Intl. Conf. on Tools with Articial Intelligence, pages 391{ 398, Chicago, IL, USA, November 1999.

....The word agent is rarely found in this overview, even though the ubiquitous use of this word would allow to classify all the presented systems as such. This is a deliberate choice. 16] is an extensive online bibliography on agents. This overview is a revised version of the one contained in [43]. 2 Applications of Personalization Applications are coarsely divided into two areas: personalized access to some resources, and approaches involving ltering. 2.1 Personalized Access As the Web continues to gain popularity, it is not surprising there do exist commercial providers for ....

....9 www.cs.umn.edu Research GroupLens 2. 2 Filtering and Rating 9 Information sources: Quality assessments of articles are based on explicit feedback and the time a user spent on a page (an approach also investigated in [36, 38, 40] and, with some modi cations, is also implemented in [43] where it is discussed in some depth. Collaborative vs. individual: GroupLens is not suited for individual personalization (and can therefore be seen as a recommendation service as discussed below) A more recent system is Alipes [58] which allows for explicitly modeling disinterest in a ....

A. Pretschner. Ontology Based Personalized Search. Master 's thesis, The University of Kansas, Lawrence, KS, 1999. www4.in.tum.de/pretschn/papers/kuthesis.ps.gz.

....www.yahoo.com, www.firefly.net, and www.pointcast.com, respectively) and filtering rating systems: electronic newspapers (e.g. Wall Street Journal or FishWrap [4] at www.wsj.com and www.sfgate.com, respectively) Usenet news filtering, recommendation services (browsing, navigation) and search. [20] describes about 45 personalization systems and contains a detailed bibliography. To the authors knowledge, SmartPush [9] is currently the only system to store profiles as concept hierarchies. These are much smaller (40 600 nodes) and weight adjustments are done with respect to data that ....

....[26] and ifWeb [1] aim at personalized search and navigation support. Syskill and Webert [19] is another example of a personalized recommendation service. InformationLens [13] is a tool for filtering and ranking e mails. Finally, 27] describes a system for expertise location (JAVA source codes) [20] contains a thorough discussion of these and other systems. Implicit rating and filtering are, among others, discussed in [17] and [18] 3. Determining the content of documents User interests are inferred by analyzing the web pages the user visits. For this purpose, it is necessary to determine ....

[Article contains additional citation context not shown here]

A. Pretschner. Ontology Based Personalized Search. Master 's thesis, The University of Kansas, Lawrence, KS, 1999. www4.in.tum.de/pretschn/papers/kuthesis.ps.gz.

No context found.

A. Pretschner. "Ontology Based Personalized Search". Master 's thesis, Department of Electrical Engineering and Computer Science - University of Kansas, 1999.

No context found.

Pretschner,A.,Gauch,S., Ontology Based Personalized Search, In Proc. 11th Intl. Conf. on Tools with Artificial Intelligence,(1999) 253

No context found.

A. Pretschner. "Ontology Based Personalized Search". Master 's thesis, Department of Electrical Engineering and Computer Science - University of Kansas, 1999.

No context found.

Pretschner A. Ontology based personalized search. Master's thesis, The University of Kansas, Lawrence, KS, 1999

No context found.

A. Pretschner. Ontology Based Personalized Search. MSc. Thesis. Kansas Univ., USA, 1999.

No context found.

A. Pretschner and S. Gauch. Ontology Based Personalized Search. In Proc. 11th IEEE Intl. Conf. on Tools with Artificial Intelligence (ICTAI'99), pages 391--398, 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC