Using Lucene/Solr in E-Learning to Implement Conceptual Extraction and Integration of Multimedia Documents
Résumé
The purpose of this work was to provide relevant integrated multimedia documents to learners. It was situated in an E-Learning environment. Concept extraction methodology based on Lucene and Solr was reviewed and adapted to learning system. Retrieval of relevant materials from a domain was implemented after relevant information was organized and related. Lucene was a key concept that helped us to relate information for providing the relevancy of lessons to the learner. It generated a learner specific e-Learning content by comparing the concepts with vector space model similarity measures. Based on the proposal, an apache projects combining Lucene and Solr were used, and vector space model and Boolean model were applied, for the searching and indexing of learner's query term from multimedia documents. Vector space model was used in information filtering, information retrieval, indexing and relevancy rankings. Vector space model allowed computing a continuous degree of similarity between queries and documents. The work selected query items which were equally weighted compared with Boolean model. The system on lucene/solr allowed multi-media document(s) to be retrieved and provided learner access to multimedia information. The resulting output to queries confirmed that Boolean model has some difficulties that included the fact that (a) it cannot allow result to be effectively ranked (b) it is difficult to rank output, data retrieval rather than information retrieval. Vector space model retrieved fewer documents compared to too many in Boolean model. Retrieving similar documents and matching all the documents so as to produce relevant document is not possible in Boolean model. The results of the proposed e-Learning system under the design of Apache projects similarity measure show a significant increase in performance and accuracy under different conditions. The assessment of the comparative analysis, showed the difference in performance of our proposed method over other methods.
Domaines
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |