The paper summarizes our research in the area of unsupervised categorization of Wikipedia articles. As a practical result of our research, we present an application of spectral clustering algorithm used for grouping Wikipedia search results. The main contribution of the paper is a representation method for Wikipedia articles that has been based on combination of words and links and used for categoriation of search results in this repository. We evaluate the proposed approach with Primary Component projections and show, on the test data, how usage of cosine transformation to create combined representations influence data variability. On sample test datasets, we also show how combined representation improves the data separation that increases overall results of data categorization. To implement the system, we review the main spectral clustering methods and we test their usability for text categorization. We give a brief description of the system architecture that groups online Wikipedia articles retrieved with user-specified keywords. Using the system, we show how clustering increases information retrieval effectiveness for Wikipedia data repository.
Authors
Additional information
- DOI
- Digital Object Identifier link open in new tab 10.3389/frobt.2016.00078
- Category
- Publikacja w czasopiśmie
- Type
- artykuły w czasopismach
- Language
- angielski
- Publication year
- 2017
