Ticket Tracking Chatbot Based On Software Engineer

Infomation Technology

It is envisaged that chatbots would fundamentally alter the field of software engineering by enabling practitioners to communicate with various services through natural language and ask questions on their software projects. Natural Language Understanding (NLU) is the core technology that powers chatbots and allows them to comprehend natural language input. Lately, a lot of NLU platforms were offered as a ready-made NLU element for chatbots; yet, choosing the ideal NLU for Software The challenge of building chatbots is yet unsolved. Thus, we assess four of the most popular NLUs in this paper: IBM Watson, Google Dialog flow, Rasa, and Which NLU should be utilized in chatbots based on software engineering will be clarified by Microsoft LUIS. We specifically look into how well the NLUs perform in extracting entities, confidence score stability, and intent classification. In order to assess the NLUs, we make use of two datasets that represent two typical tasks carried out by practitioners of software engineering: a chatbot for software repository inquiries the activity of posting queries about development on Q&A sites (like Stack Overflow).Based on our research, IBM Watson is the NLU with the best performance across the three dimensions (entity extraction, confidence scores, and intents categorization). The results for each individual component, however, indicate that Rasa leads in confidence scores with a median confidence score greater than 0.91, while IBM Watson performs best in intents categorization with an F1-measure>84%. .. Additionally, our data demonstrate that all NLUs—aside from Dialog flow— generally offer reliable confidence scores. For entity extraction, Microsoft LUIS and IBM Watson outperform other NLUs in the two SE tasks. Our results provide guidance to software engineering practitioners when deciding which NLU to use in their chatbots.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.