Semantic Graph Knowledge Repository Designed For KID Cognitive Model Applied In Retail Business

Retail companies can no longer afford to make decisions based on gut feeling.Now it is a time that companies must take advantage of all available mountains of data to do advanced data analytics which can capture value and delivery useful insights to guide their decision making to occupy a chunk of market share against competitors.However,retail companies still have difficulties in processing large amount of data to make better predictions and marketing strategies for their survival.As marketing becomes more customer-centric,more analytics and research should be done on customer purchase behavior.Based on how human process outside information and absorbing traditional big data analytic method which divides big data into pieces of small data,KID(Data-Information-Knowledge)model which is a generic from data to knowledge conitive model was proposed by Huang and Sato to support retail data analytics.Streaming data is interpreted into meaningful information by prior or expert knowledge stored in the knowledge repository of KID model.Then,this meaningful information is absorbed or assimilated into existing knowledge repository.Knowledge will be added,modified or updated according to this meaningful information based on knowledge fusion.The core part of KID model is the knowledge repository which should be designed to support interpreting streaming data into meaningful information,assimilating meaningful information and updating knowledge.A retail semantic graph knowledge repository embedding a pragmatic rule-based knowledge discovery engine is proposed and developed for KID model by the integration of Gruninger and Fox method,Neo4j graph database,retail ontology designed by Maryam Fazel Zarandi and Jess rule engine to interpret retail data into information,assimilate information into graph knowledge repository to update knowledge and deduce answers to retail queries based on the general knowledge of retail business.A pragmatic rule-based knowledge discovery engine including working memory,an inference engine and a rule base which has forward chaining rules and backward chaining rules is embedded in graph knowledge repository to provide the deductive reasoning ability.The backward reasoning method is used to gather knowledge from graph knowledge repository based on Neo4j graph database and assert them as facts into working memory of rule engine.The forward reasoning method is responsible for answer a given query.We also develop semantic graph knowledge repository based knowledge discovery system to make the proposed retail semantic graph knowledge repository applicable in real retail business world.A customer value assessment case study by using Recency-Frenquecy-Monetary(RFM)analytic model and K-means algorithms is given to demonstrate the proposed semantic graph knowledge repository and KID model based knowledge discovery system.