An Approach for Building a Semi-automatic Online Consultancy System
Nguyen Thai-Nghe, Quoc Dinh Truong · 2015
This study proposes an approach for building a semi-automatic consultancy (question-answering) system via mobile/Internet networks. This approach is a combination of natural language (text) processing and machine learning method. For building the system, at first, we need to build modules for sending and receiving messages via SMS/Email/Webpage. These modules are used for users to send/receive their questions that need to be consulted. While waiting for answering from the system, the user will be recommended similar questions which have been answered in the past by using Cosine similarity. Next, a message classification module is built using a combination of text processing (e.g., word segmentation, stop word deletion) and machine learning method (e.g., SVM). Finally, a whole web-based system is conducted to integrate these modules. The proposed approach is applied for a case study of consulting on Vietnam National Entrance Test, which is an important test for the pupils to get into universities. Initial results show that the system can automatically classify the questions at 82.33% of accuracy, thus, this approach could be promising for (semi) automatic online consultancy systems.