Papers101: Supporting the Discovery Process in the Literature Review Workflow for Novice Researchers
To appear at the Proceedings of IEEE Pacific Visualization 2021.
An overview of Papers 101, an interactive system that accelerates the discovery of literature for novice researchers. (A1) Keyword Search Box and Keyword List which receives search keywords and list them;(A2) Recommended keywords which recommend other relevant key-words to current search query;(B1) Ranking view which visualizes the ranking of the papers;(B2) History view which visualizes up to five previous rankings histories in parallel coordinates;(C) Citation relationship which enables users to validate the cohesiveness of the seed paper set;(D) Paper detail which provides additional information of the selected paper; and (E) Seed paper list which shows a list of seed papers selected by users.
A literature review is a critical, but daunting task for novice researchers. Once the research topic is decided, the literature review process should follow; researchers need to review papers from various perspectives and synthesize the knowledge. However, novice researchers often face difficulties even at the initial phase, browsing relevant work from academic databases and choosing must-read items among them. This can be for many reasons such as vague ideas about appropriate search terms and less knowledge on exploiting academic metadata.
What further intensifies the problem is that the discovery of papers is not a single task but a process. For example, novice researchers who are less knowledgeable about terminologies in a certain domain often attempt to obtain a list of candidate papers by searching with common, broad keywords. Through the iterative queries, they eventually learn more suitable academic terms to use for the following search. Thus, until they reach a final understanding, various bottlenecks of novice researchers occur repeatedly and simultaneously. If individual challenges are handled by separate tools, it will increase a burden to integrate and switch between multiple contexts. The absence of integrated tools is a problem that is pointed out among researchers who are not novices, and it should be particularly addressed in systems intended for beginners.
We performed a formative study and design iterations with researchers from varying level of experience. As a result, we identified not only the unique tasks and needs that exist in the process, but also the empirical heuristics that can resolve it. This lead to the design of Papers101, an interactive system that supports novice researchers to discover papers relevant to their research topic. Papers101 provides opinionated perspectives on the selection of important metadata according to the stage and purpose within the discovery process. It also visualizes how the priority among papers are composed of those metadata and how it has changed over the users' path of knowledge discovery. In addition, Papers101 provides ways to validate and refine the current search query so that the users can be guided beyond their local knowledge and interests.