Path Ranking Algorithm

In the next lab meeting on July 10, Maryam Siahbani is going to talk about Ni Lao and William Cohen’s paper on the “Path Ranking Algorithm” (PRA) and some applications of this method. PRA uses random walks to generate relational features on graph data, and combine them with a logistic regression model. This approach has been used to solve efficiency issue of relational retrieval and relational learning in real-world applications with large-scale data (e.g. inference task in a large knowledge base).

The original paper is Relational Retrieval using a combination of path-constrained random walks.

There have been two papers on its application as well.

• Ni Lao, Tom Mitchell and William Cohen. Random Walk Inference and Learning in A Large Scale Knowledge Base
• Ni Lao, Amarnag Subramanya, Fernando Pereira, and William Cohen. Reading The Web with Learned Syntactic-Semantic Inference Rules

The lab meeting will start at 1230 hours at TASC1 9408.