Software developers often need to read code snippets that are dispersed among different documentation, e.g., Q&A posts, to reuse APIs to complete certain tasks. These code snippets are often surrounded by lengthy context text which are used to describe the functions of code snippets. It will be helpful for code comprehension if we can align a code snippet with its description. In this paper, we propose an approach to extracting code-relevant sentences from its context text. To quantify the relevance between code line and natural language sentence, we represent them with structure trees and calculate their structural similarity. We conduct two experiments to evaluate our approach. In Experiment I, the results show that our approach achieves 83.5% precision and 80.1% recall in aligning Lucene code snippets and corresponding comments. Our approach achieves 27.6% ~ 40.2% improvement in precision compared with existing method, and 33.8% ~ 39.7% improvement in recall. In Experiment II, the results show that our approach achieves 66.4% ~ 93.9% precision to extract code-relevant sentences.

Different kinds of software documents are produced in the life cycle of a software project, such as Bug Reports, Mail Lists, etc. These documents have close relationship with source code, but it is difficult to recover their traceability relationship. In this paper, we conduct an exploratory study on codes in a software project's heterogeneous documents, so that we can give some hints for traceability recovery from software documents to source code. We select a famous open source software project, Lucene, as sample, and collect its four kinds of software documents, including Bug Reports, Mail Lists, Stack Overflow Q&A Documents and Blogs. On this basis, we analyze these heterogeneous documents to answer the following questions: How much code is there in different kinds of documents? What APIs do these documents focus on? How many documents are relevant to the same APIs? Based on the study, we give 3 hints for recovering the traceability from software heterogeneous documents to source code.

Recovering traceability links between source code and software document can be very helpful for Software Maintenance and Software Reuse. Existing work has already achieved good results in extracting code elements (classes, methods, etc.) from software documents. However, it will lead to a lot of noise links if we link a document to all the code elements existing in it. In this paper, we propose an approach to identify the contextual code elements and the salient code elements in a software document, then we can weight the traceability links between source code and software document so that those noise traceability links can be filtered effectively. We measure the saliency of each code element in a document with four kinds of document-related features and three kinds of code-related features, and we adopt TransR-based code embedding technology to evaluate the distance between code elements. In the experiments, we get a precision of 70.7% in recognizing salient code elements of StackOverflow answer documents, which is more than 12% improvement compared with Rigby's work. At the same time, we can filter about 56.5%~69.3% noise traceability links compared with the RecoDoc approach. It will improve the quality of traceability links between source code and related software documents.