ScienScan - An Efficient Visualization and Browsing Tool for Academic Search

Daniil Mirylenka, Andrea Passerini

Introduction

In the realm of academic research, "exploratory search" represents a critical phase where scholars attempt to map out unfamiliar scientific landscapes. Traditional search engines like Google Scholar or Microsoft Academic Search present results in linear, unstructured lists. While efficient for finding specific known papers, these interfaces are often "tedious and even frustrating" for users seeking a semantic overview of a field. ScienScan addresses this by transforming flat search results into interactive, hierarchical topic maps.

Methodology: From Flat Lists to Topic Graphs

ScienScan functions as a post-processing layer that enhances existing search engine outputs. The system follows a structured five-step pipeline:

  1. Data Retrieval: The system fetches the top 100 results (titles and abstracts) via the Microsoft Academic Search API.
  2. Topic Annotation: Using Wikipedia Miner, the text is mapped to specific Wikipedia articles, which serve as human-labeled semantic anchors.
  3. Graph Construction: By identifying parent categories within Wikipedia's taxonomy, the tool builds a Directed Acyclic Graph (DAG) of related concepts.
  4. Summarization: To prevent visual clutter, a frequency-based heuristic algorithm selects the most informative subset of nodes, ensuring high coverage with minimal redundancy.
  5. Visualization: The final hierarchy is rendered using the Graphviz package.
Hierarchical topic map for 'mathematical optimization' showing how topics branch into sub-fields like quadratic and linear programming.

A key technical innovation is the "incremental growth" property. As users adjust the granularity of the map via a slider, the system ensures that smaller maps are strict subsets of larger ones. This prevents topics from appearing or disappearing erratically, maintaining the user's mental model during exploration.

User Interface and Semantic Mapping

The core strength of ScienScan lies in its use of Wikipedia as a semantic proxy. Unlike unsupervised clustering methods that often produce cryptic labels, ScienScan utilizes established Wikipedia categories, ensuring that every node in the visualization is immediately recognizable to the user.

The ScienScan interface featuring the search query 'probabilistic topic models', a granularity slider, and the resulting interactive topic map.

The interface is highly interactive. Users can click on specific nodes (e.g., "Topic model") to instantly filter the underlying list of papers. Furthermore, the tool employs proportional font sizing: the larger the label, the more papers that topic covers. This creates a visual "heat map" of research density within the search results.

Insights and Future Directions

ScienScan demonstrates that real-time, on-the-fly visualization of academic metadata is feasible and provides a significant navigational advantage. By allowing users to "tune" the granularity of their search, the tool bridges the gap between high-level overviews and deep-dive technical reviews.

Future work aims to move beyond heuristic summarization. The authors are exploring structured-output prediction models to create even more logical and intuitive taxonomies. While currently limited by API constraints (100 results per query) and the maturity of specific Wikipedia sub-categories, ScienScan represents a significant step toward more intelligent, semantically-aware academic search tools.

Summary updated

Abstract

In this paper we present ScienScan – a browsing and visualization tool for academic search. The tool operates in real time by post-processing the query results returned by an academic search engine. ScienScan discovers topics in the search results and summarizes them in the form of a concise hierarchical topic map. The produced topical summary informatively represents the results in a visual way and provides an additional filtering control. We demonstrate the operation of ScienScan deploying it on top of the search API of Microsoft Academic Search.
BibTeX
@inproceedings{mirylenka2013scienscan,
  author = {Daniil Mirylenka and Andrea Passerini},
  title = {{ScienScan - An Efficient Visualization and Browsing Tool for Academic Search}},
  booktitle = {ECML/PKDD},
  year = {2013},
  doi = {10.1007/978-3-642-40994-3\_51},
}
Figure