Graph Analytics and Platforms - overview
This effort focuses on developing core Research capabilities at platform layer. The two main areas of work are:
Temporal Graphs: The focus of this work is to enable new age data platforms like Titan and APACHE SPARK to handle temporally changing data and maintain complete provenance and audit trails. The technical challenges range from developing strategies to store such data, to effectively support simple temporal queries and to develop novel temporal pattern mining algorithms on the enhanced platforms.
Tools to Convert Legacy Data to Property Graphs: This effort focuses on developing automated tools to port data in traditional data stored to property graph stores. The tools take into account the query workload and data characteristics to generate "correct shape of the graph" which will provide better (at least comparable) performance to existing stores. We envision to develop a whole suite of analytics and optimization tools which will help to load the data and continuously tweak the graph to reflect more recent workload.
We are looking at developing large scale graph analytics algorithms which are compatible with Titan and SPARK. Some of the current problems we are looking at include contextual recommendations, entity resolution and graph visualization. We are also exploring use to Spatial-Temporal data and how it can impact the recommendations. The other area which we are actively working on is to integrate Text and Graph data to support user driver exploratory Information Retrieval.