![]() | Only 14 pages are availabe for public view |
Abstract The demand of knowledge has been increasing during the 21st century and knowledge seeking is always a challenging task for all organizations and professionals. With the exis- tence of the internet, online knowledge communities are built for knowledge seeking and sharing between individuals across time and space. A huge number of questions are posted over the online communities on a daily basis. The questions may face two main challenges: the {uFB01}rst is a long waiting time for a response and the second is low quality answers. In this thesis we a provide a framework that is capable of routing the new questions to the expert users who have the expertise to give a reasonable answer in a suitable time frame. In our work, we proposed a question routing technique in community question answer- ing based on a deep learning technique called deep semantic similarity model (DSSM). The proposed technique (QR-DSSM) captures the semantic similarity between the posted question and the community users and it ranks the users{u2019} pro{uFB01}les based on the similarity scores. QR-DSSM adopted the deep architecture in order to enhance the semantic structure extraction from the posted questions and the users pro{uFB01}les through using multiple non- linear hidden representation layers. QR-DSSM were able to extract more sophisticated semantic structures from the questions and the pro{uFB01}les. We performed extensive experiments to compare our proposed question routing tech- nique to the currently existing question routing frameworks |