Replication Package of API-related Developer Needs in Stack Overflow

Empirical Study

• 266 API related posts among 500 top voted posts. This csv file is the result of posts analysis. In the csv files, the column "url" represents Stack Overflow url for posts, "title" represents the title of corresponding post, "annotator_1" represents the annotation of the first annotator, "annotator_2" represents the annotator_1 of the second annotator, "arbitration" represents the arbitration result.

• developer need and relevant information analysis of 266 posts. We derived a fine-grained taxonomy through an empirical study of 266 API related SO posts from top 500 voted SO posts. Then discussed and summarized 8 developer need types and 17 relevant information types. This folder contains specific analysis of the above types. In the "relevant information annotation.json", key "id" represent the represents post ID in Stack Overflow, "label for title" contains several lists in which first items are sentences, second items are developer need types, "label for body" is similar to "label for title", "title" and "body" are the Stack Overflow question title and body. In the "456 developer need.json", key "id" represents the represents post ID in Stack Overflow, "developer need type" represents developer need types, "sentences" represents the sentences in this developer need.

• developer need and relevant information analysis of 34 posts. This folder contains the results of two annotators which contain questions from the 266 questions which contain five questions for each developer need type. In the csv files, the column "post_id" represents post ID in Stack Overflow, "text" represents the current sentence of corresponding post, the other columns are developer need types and relevant information types.

• developer need instances classification. This file contains the group result of developer need sentences. In the json file, key "post_id" represents post ID in Stack Overflow, "developer need type" represents the developer need types, "annotation_1" represents the sentences group result of the first annotator, "annotation_2" represents the sentences group result of the second annotator, "kappa" represent the kappa value.

API Identification

• code for API identification. This folder contains codes for API identification. Folder "text_based" contains scripts for extracting API from text. Folder "text_based" contains scripts for extracting API from code snippets, stack traces. Folder "combined" combined all extracting ways. We hide some Python libraries which released by ourselves in these scripts for reasons of anonymity. We plan to release our code as a python library to provide this functionality.

Annotated Data Expansion

• annotated data expansion. This folder contains 17 files corresponding to 17 relevant information types. In each csv file, the column "url" represents Stack Overflow url for posts, "sentence" represents the current sentence for annotating, "title" and "body" are the Stack Overflow question title and body, "arbitration" represents the arbitration result.

Relavant Question Retrieval

• relevant_question_retrieval. This folder contains 4 files. One is about the arbitration of experiment, the other three are experiment results. In file experiment_data_100.csv, the column "url" represents Stack Overflow url for posts,"title" and "body" are the Stack Overflow question title and body, "arbitration" represents the arbitration result. In other three files, the column "id" represents Stack Overflow post id for query,"query" represents query, "related_url" represents the url of retrieval result, "title" represents the title of retrieval result,"arbitration" represents the arbitration result.

It is hereby stated that due to anonymity, some information is hidden and will be disclosed in the future