User-Interactive Machine Learning Model for Identifying Structural Relationships of Code Features

  • 2019-07-18 16:18:51
  • Ankit Gupta, Kartik Chugh, Andrea Solis, Thomas LaToza
  • 11

Abstract

Traditional machine learning based intelligent systems assist users bylearning patterns in data and making recommendations. However, these systemsare limited in that the user has little means of understanding the rationalebehind the systems suggestions, communicating their own understanding ofpatterns, or correcting system behavior. In this project, we outline a modelfor intelligent software based on a human computer feedback loop. The MachineLearning (ML) systems recommendations are reviewed by the user, and in turn,this information shapes the systems decision making. Our model was applied todeveloping an HTML editor that integrates ML with user interaction to ascertainstructural relationships between HTML document features and apply them for codecompletion. The editor utilizes the ID3 algorithm to build decision trees,sequences of rules for predicting code the user will type. The editor displaysthe decision trees rules in the Interactive Rules Interface System (IRIS),which allows developers to prioritize, modify, or delete them. Theseinteractions alter the data processed by ID3, providing the developer somecontrol over the autocomplete system. Validation indicates that, absent userinteraction, the ML model is able to predict tags with 78.4 percent accuracy,attributes with 62.9 percent accuracy, and values with 12.8 percent accuracy.Based off of the results of the user study, user interaction with the rulesinterface corrects feature relationships missed or mistaken by the automatedprocess, enhancing autocomplete accuracy and developer productivity.Additionally, interaction is proven to help developers work with greaterawareness of code patterns. Our research demonstrates the viability of asoftware integration of machine intelligence with human feedback.

 

Quick Read (beta)

loading the full paper ...