API Requirements for Dynamic Graph Prediction

Gallagher, B ; Eliassi-Rad, T

Lawrence Livermore National Laboratory

关键词: Programming;    Learning;    Algorithms;    97 Mathematical Methods And Computing;    Forecasting;   

DOI  :  10.2172/1036864
RP-ID  :  UCRL-TR-225656
RP-ID  :  W-7405-ENG-48
RP-ID  :  1036864

美国|英语

来源: UNT Digital Library

【 摘 要 】

Given a large-scale time-evolving multi-modal and multi-relational complex network (a.k.a., a large-scale dynamic semantic graph), we want to implement algorithms that discover patterns of activities on the graph and learn predictive models of those discovered patterns. This document outlines the application programming interface (API) requirements for fast prototyping of feature extraction, learning, and prediction algorithms on large dynamic semantic graphs. Since our algorithms must operate on large-scale dynamic semantic graphs, we have chosen to use the graph API developed in the CASC Complex Networks Project. This API is supported on the back end by a semantic graph database (developed by Scott Kohn and his team). The advantages of using this API are (i) we have full-control of its development and (ii) the current API meets almost all of the requirements outlined in this document.