Introduction to Interest-based Selecter
Interest-based Selecter 2.0 is developed by International WIC Institute, Beijing University of Technology, China. It is compatible with LarKC version 2.0.
Plugin authors (Yan Wang, Yi Zeng, Xu Ren)
Please contact Yi Zeng (yizeng@bjut.edu.cn) for the use of this plugin and collaborations.
Test Suite
Input：N-triple format Set Of Statements from Identifiers
Ouptut: N-triple format Set Of Statements as selected sub dataset
Where to Download the plugin
http://www.wici-lab.org/wici/larkc/plugins/InterestBasedSelecter.rar
Description to the Plugin
Selection method:
This plugin first calculate and obtain Top-K user interests from a set of user related triples, then it selects sub datasets that contains Top-K user interests. The selected triples are passed to LarKC data layer.
Two methods to obtain user interests:
Read from an existing user interest profile (Wrapped as a Named Graph, and passed from an identifier, the name of the interest graph should include "interest.larkc", for example: http://twitter.com/Frankvanhamele/publishedInfo.larkc). This function is migrated from Interests-based Selector v1.0.
- Real time calculation of user interests based on set of statements (related to a specific user) passed from an Identifier. The name of the graph should include "publishedInfo.larkc".
Contracts to interests related data from identifier:
- The format of the named graph : interest sources (address+username) + interest type
- There are 3 types of interests, we use the following to denote them:
(1) "interest.larkc" : interest graph that is calculated offline and represented by using FOAF or e-FOAF: interest vocabulary.
(2) "publishedInfo.lark": User related text contents, formated as set of statments, and are used for calculation of user interests in real time.
(3) "?LocationDescriptionFriend.larkc": user's relatively static interests, such as registration information, etc.