The increased use of online collaboration tools and services has resulted in the accumulation of large volumes of information pertaining to software development and related activities. Such sources of raw/unstructured knowledge include but are not limited to:

• The platforms such as stackoverflow.com that allow professional developers to discuss a variety of software development issues.

• GitHub allows OSS communities to collaboratively develop industrial-strength software.

• Software project bug reports repositories.

• Vendor specific documentation portals such as MSDN, IBM Developer Network, etc. that describe best practices as well as usage patterns for a variety of technologies developed by those vendors. Such technologies often serve as the core fabric for building customized products and solutions for a variety of domains.

Challenge: However, the challenge is that such raw content has to be first transformed into a form that is suitable for automatic knowledge analysis, reasoning, and recommendations.

Objective: The objective of this Vritika program is to train the students to develop a set of novel algorithms and tools for:

1. Automatic extraction of raw data from different sources highlighted in the preceding section.

2. Automatic transformation of raw data into a structured form which is suitable for consumption by various Machine Learning (ML) algorithms that perform automatic analysis/reasoning on such data for answering domain-specific queries.

3. Developing domain-specific knowledge models that perform automatic analysis/reasoning on such data for answering blockchain system-specific queries.

4. Developing suitable consumer-facing service APIs that will allow 3rd parties to build intelligent services and tools that exploit our knowledge warehouse and models for blockchain-based systems.

Focus: The programme focuses on the implementation of various Software Engineering, Information Retrieval (IR) and ML techniques in the domain of Software Engineering with the motive of developing automated solutions for defect (or bug) prediction and recommendation in blockchain-based systems.

Deliverables: The projects undertaken by the students in Vritika will lead to the following deliverables:

• Knowledge Warehouse (KW) containing all the relevant information in a structured format.

  • Dataset of features extracted from Blockchain-based systems

  • Dataset containing the details of bugs associate with the considered Blockchain-based systems

  • ML-trained models on the above features

• Discovery of interesting patterns that exist in the data collected in the KW

• These patterns when used with various ML techniques could be used to perform the various subtasks like Bug localization, Source Code Recommendation, etc.

• User Interface to help the user query the required information in a systematic manner.