Search

With the increasing number of web sites, the time spent by users reviewing the results also increases. In addition, the nature of web content is semantically heterogeneous and oriented to people who will be able to understand it. Frequently the results from search engines do not correspond to the expected topic. One approach to improve the results is to match the content of the web pages with a formal vocabulary on the topic (ontology) and with the informal vocabulary (common terms of the topic but not in the ontology). This paper describes a web search method that takes advantage of ontologies to reduce the search area of certain topics. With this approach the relevance of search engine results is enhanced by filtering the content through the integration of domain ontologies, the WordNet thesaurus, and a hierarchical similarity measure. Thus, the improvement on the relevance of results reduces the time required to review such results. [ABSTRACT FROM AUTHOR]

Improvement of the interfaces and search capacities of the opacs continues to be one of the main challenges for libraries, especially with regard to searching by subject. The visual interfaces can help with retrieval. This paper explores the combination of co-term analysis techniques with social networks for the generation of subject maps of collections. The main conclusion is that the methodology applied is valid, and that the maps could be used for browsing and subject retrieval. It would also be useful for improving the quality of indexing. [ABSTRACT FROM AUTHOR]

This paper offers an approach to the implications of digitising a radio station's audio archive, based on a case study of the public broadcaster Catalunya Ràdio. Beyond the unquestionable importance of factors such as technical equipment, software or standards of compression, and storage and indexing, the conversion of an audio file into binary code has other implications that are briefly reviewed. The essential parts of this process can be summarised as follows: changes in professional profiles and production routines, the integration of languages, and the possible universality of documentary resources. [ABSTRACT FROM AUTHOR]

This paper shows how the individual study of different basic attributes from a web resource is not enough to infer different positioning strategies for search engines. The fundamental question is about the relation between the different elements that compose the page and the importance that each of them contributes to the final positioning. As an alternative to this problem, we propose the use of inductive learning techniques, more concretely, sort trees. Results are reflected in two experiments, the outcome of two different learning algorithms. As a final result, we conclude that the application of this technique can be a very promising starting point for optimising web positioning. [ABSTRACT FROM AUTHOR]

This paper analyses the different positioning strategies according to their ability to obtain an optimum recall level. The analysed positioning strategies are the search for terms in the title, in the url, in the links, or in the Meta and Head tags. Our analysis is restricted to non-commercial pages--commercial pages are avoided because of their policies for achieving greater popularity by means of web optimisation-. The specificity level of the query is taken into account. We submitted thirty queries to four search engines (AskJeeves, Google, YahooSearch, and MSN); then we judged by hand the relevance of the first twenty hits of each query. Our analysis yields the following findings: the utility of the title strategy; the need for analysing the data by function of the type of query; and the suitability of designing new retrieval measurements. These measurements should take into consideration the relevant documents without the strategy, as well as those non-relevant documents with it. [ABSTRACT FROM AUTHOR]