Human resource matching through query augmentation for improving search context

• Main Author: Tran, Huy
• Format: Others
• Language: English
• Published: Linköpings universitet, Databas och informationsteknik 2016
• Subjects: Computer and Information Sciences; Data- och informationsvetenskap
• Online Access: http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-129571

The objective with the thesis is to research how to match a company's human resources with job assignments received from clients. A common problem is the difficulty for computers to distinguish what semantic context a word is in. This means that for words with multiple interpretations it is hard to determine which meaning is the correct meaning in a given context. The proposed solution is to use ontologies to implement a query augmentation that will improve defining the context through users adding suggestions of relevant words. The intuition is that by incrementally adding words, the context narrows, making it easier to search for any consultant matching a specific assignment. The query augmentation will then manifest in a web application created in NodeJS and AngularJS. The experiments will then measure, based on \emph{precision}, \emph{recall} and \emph{f-measure}, the performance of the query augmentation. The thesis will also look into how to store document-based résumés, .docx file-format, and properly enable querying over the database of résumés. The Apache based frameworks Solr and Lucene, with its inverted indexing and support for HTTP requests, are used in this thesis to solve this problem. Looking at the results, the query augmentation was indicated of having somewhat too strict restrictions for which the reason is that it only permits \emph{AND} conditions. With that said, the query augmentation was able to narrow down the search context. Future work would include adding additional query conditions and expand the visualization of the query augmentation.