Research On Product Quality Event Monitoring And Tracing Based On Hierarchical Dirichlet Process

With the vigorous development of e-commerce,product quality feedback data grows explosively all around the network,which seriously exacerbates the uncertainty about quality and safety.This is illustrated by frequent network product quality incidents in recent years.So,it is a question of great theoretical and practical significance to discover key information about product quality from the massive network information,which makes for improving the efficiency of quality control process.In respect of the issues above,the methods of product quality event monitoring with Hierarchical Dirichlet Processes topic model and product quality event tracking based on ontology are studied.The main work done in this paper is as follows.(1)Method researching on product quality event monitoring based on Hierarchical Dirichlet Processes topic model.Considering the feature sparseness of product corpus and topic mining effect,the topic mining algorithm based on the Hierarchical Dirichlet Process is applied for modeling,then take it to carry out to topic clustering.Secondly,analyze the evolution trend of topic intensity and word contents by timescales,and then monitor changes of product quality characteristics.Lastly,referring to overmuch redundant words existing in topic clustering results,add topic words sort processing measure based on the original topic mining method,and it can reduce the probability of redundant words,also make product topic characteristics more precise.(2)Modeling of product quality event traceability based on ontology.Through research and analysis of product quality event knowledge,the generalized event types,event elements,and semantic relationships between events are abstracted in the area of product quality events,and the domain abstraction framework and trigger word-based event class division method are used to extract knowledge.At the same time,according to the characteristics of quality event knowledge,the product quality event representation ontology model is constructed from four aspects: event elements,event text features,event relationships,and product components.The model can be used to query event semantic information.Lastly,traceability directed graphs are used to build traceability chains for realizing event flow traceability.(3)Case verification using the product of Samsung Galaxy note7 as the object.Adopt the above monitoring method to carry out topic evolution analysis of the product dataset,and then monitoring product quality event.Then taking the battery explosion event of note7 as the research object,denote semantic information and association relation among event classes,then trace the evolution of events with the created event ontology.In the paper,the method of quality information monitoring based on the Hierarchical Dirichlet Process is studied,and the ontology model towards product quality event is built.Finally,it is verified to be effective through examples of major product quality incidents in recent years.