A hierarchical multi-relational clustering algorithm based on modal logic
2011
For datasets contained multi interrelated tables, multi-relational clustering divides target objects into clusters according to their attributes and features of objects related to them directly or indirectly. Due to the actual business, all target objects don't exist information in every nontarget relation, so target objects may be described by information of different order. To get information about one-to-many relationships, it is often unable to reflect original distribution of data if using statistics directly. To solve these problems, we propose a new method to model multi-relational data set based on modal logic, define distance between objects, and clustering by means of original features of all objects. Experiments indicate that our method can dispose information of different order effectively, and obtain more accurate and reasonable clustering results.
Keywords:
- Artificial intelligence
- Correlation clustering
- Cluster analysis
- Consensus clustering
- Pattern recognition
- Determining the number of clusters in a data set
- FLAME clustering
- Canopy clustering algorithm
- Computer science
- Machine learning
- CURE data clustering algorithm
- Brown clustering
- Constrained clustering
- Data mining
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
11
References
0
Citations
NaN
KQI