| Résumé : |
(Auteur) The user's need to find information on the Internet is growing, like the Internet itself is growing into an ever more tangled web of information. To find information on the Internet, search engines have to rely solely on the techniques and the mod-els developed in the information retrieval, due to the unstructured nature of the Internet. When focusing on smaller domains of the Internet. like Intranets, web-sites, and digital libraries, large collections of documents can be found. which have both a semi-structured and a multimedia character. The information contained in such document collections is often semantically related. which in total provide the conditions to successfully apply existing database techniques for modelling and querying web-data. The Webspace Method focuses on such document collections and offers a new approach for managing and querying web-based document collections, based on the principle of conceptual modelling. The conceptual schema derived in the modelling stage of the Webspace Method is called the webspace schema. The webspace schema contains concepts, which describe the information stored in the document collection at a semantical level of abstraction. A webspace consists of two levels: a document level, where the document collection is stored, and a semantical level, which consists of the concepts defined in the webspace schema. The relation between the concepts defined in the webspace schema and the content stored at the document level is exploited by the Webspace Method when querying a webspace. The schema-based approach for querying the content of a document collection, introduces a new category of search engines, which offer more precise and advanced query formulation techniques. Querying a webspace, i.e. the document collection, is based on the combination of conceptual search and existing information retrieval techniques. The Webspace Method allows a user to formulate both conceptual and multimedia constraints over the collection of documents as a whole, rather than to query the content of a single document at a time. Based on the webspace schema, a user formulates his information need by defining some conceptual constraints, extended with con-straints on the different types of multimedia that the user wants to invoke in his search assignment. Because the conceptual constraints of the query are based on the webspace schema, it is possible to combine information originally stored in several documents into the result of the query. Because of the conceptual frame-work, the webspace search engine is capable of finding information that cannot be found, using standard search engine technology. The integration with information retrieval allows the Webspace Method to combine content-based retrieval with conceptual search on limited domains of the Internet. The user can then easily integrate several multimedia (sub-)queries into the conceptual query over a webspace. Furthermore, the webspace search engine allows a user to formulate his information need in terms of concepts, as the result of a query. Rather than returning a list of URLs to relevant documents, a mate-rialised view on the webspace schema is generated, which contains the requested information. To evaluate the Webspace Method, a retrieval performance experiment is car-ried out to measure the contribution of the Webspace Method to the retrieval process, compared to a 'standard' search engine, using the same IR model. It shows that the retrieval performance, measured in terms or recall and precision, increases by a factor up to 2, when querying a document collection with the Web-space Method.
|
Ajouter le résultat dans votre panier Affiner la recherche Interroger des sources externes
