TiQi: Natural Language Interface for Querying Software Projects

Software and systems engineering projects accrue large amounts of development data including requirements, code, test cases, fault logs, versioning data, architectural documents, and design rationales. When combined with the power of software analytics this data could be used to support the decision making of project stakeholders by delivering actionable intelligence. For example, a developer might ask for a "list of all classes which are safety-related and likely to exhibit runtime faults." Results from this query could help the developer focus effort on the most critical, fault-prone parts of the code. Despite the potential benefits of utilizing project data to address such questions, participants at a recent NSF-hosted meeting on Strategic Priorities for Software Development identified the inability to use project data effectively as a critical issue retarding time-to-market competitiveness and affecting the quality and safety of delivered products.

Our work in this space explores visual and natural language query mechanisms. The Visual Trace Modeling Language allows users to express trace queries using a UML profile, while TiQi supports natural language queries such as "Show me a list of high-priority requirements which have been implemented in code this past week".

While we have tested TiQi against close to 300 queries, it is still a work in progress. We have therefore released a limited version of TiQi here which interprets and then executes only a small set of queries. Our next release, anticipated for Spring 2016, will allow open queries.

VTML and TiQi publications include the following:

• Piotr Pruski, Sugandha Lohar, William Goss, Alexander Rasin, Jane Cleland-Huang: TiQi: answering unstructured natural language trace queries. Requir. Eng. 20(3): 215-232 (2015)
• Piotr Pruski, Sugandha Lohar, Rundale Aquanette, Greg Ott, Sorawit Amornborvornwong, Alexander Rasin, Jane Cleland-Huang: TiQi: Towards natural language trace queries. RE 2014: 123-132
• Patrick Maeder, Jane Cleland-Huang: A visual language for modeling and executing traceability queries. Software and System Modeling 12(3): 537-553 (2013)
• Patrick Maeder, Jane Cleland-Huang: A Visual Traceability Modeling Language. MoDELS (1) 2010: 226-240

Current Researchers and Collaborators

• Dr. Jane Cleland-Huang, SAREC Director, Professor DePaul University
• Dr. Alex Rasin, Assistant Professor DePaul University
• Sugandha Lohar, PhD Student
• William Goss, Research Assistant (Graduated with BS in 2014)
• Marlon