Wiki Gardening Engine
- Mentor(s)
- Details
Overview
As a knowledge management software, it is critical for XWiki to help its users to better organize the information that is put on the wiki.
One frequent problem of wikis is that content is being put in wikis but can not be capitalized/used because it's difficult to retrieve once put in the wiki (because not properly organised, or because search engines are not performant enough).
Usually, in all communities that collaborate around creating a knowledge base, there are a few people (the "gardeners") that watch over the good organisation of the wiki, make sure that all content that is being added is not duplicate, that it is put in the proper place on the graph of knowledge, that relations are properly setup with related documents, etc. The purpose of this project is to use all available language processing / artificial intelligence / machine learning tools in order to create replace the gardeners with robots.Objectives
The main goal of this project is to build a tool to help document editors to better organize their pages and the pages contents. This tool should be a combination of machine learning coupled with natural language processing. The tool should be able to propose different forms of organisation of content to the editors, which editors can apply to their content or not. Ideally, the automatic gardener would learn from the answers of the editors and improve its algorithm.
A few examples of gardening actions:
- detecting duplicate content
- proposing to update a page rather than create a new page
- inferring a structure (in the sense of the wiki's structured pages) from the created content and proposing to add this structure to a set of pages
- proposing a location for a page on the wiki (depending on its content)
- etc. The list is really open, a full analysis of the possible actions should be done by the student.
Requirements
- Good knowledge of machine learning
- Good knowledge of Java
- Basic understanding of knowledge management needs
Other resources
- Year
2019
- Status
Proposed