Enhanced Topic Modelling using Dictionary For Questions and Answers Problem
Maryamah Maryamah, Agus Zainal Arifin, Riyanarto Sarno, Rizka Wakhidatus Sholikah · 2019
Making Questions and Answers (QA) with large data and a broad context of problems can cause the desired document to sometimes be irrelevant. QA in terms of religious-social issues have a broad context, so they need to be firstly introduced to the topic. However, the questions raised in the Questions and Answers problem have short text criteria that must be correctly identified by the topic according to the relevant answers. In this paper, we proposed topic modeling for questions and answers with improved term weighting in special words in religious-social problems. The process consisted of preprocessing, making improved dictionary and modeling topic based on dictionary. The result obtained was in the form of topics from input short text which assisted in taking relevant topics, so that correct answers to the questions were obtained.