Jin Maozhong

A new clustering method for collaborative filtering

Collaborative filtering, which predicts unknown ratings by analyzing the known ratings, is regarded as one of the most successful recommender systems within the last decade. In the process of collaborative filtering, clustering methods can recognize similar users and items. Then different methods can be applied to process different clusters. By these means, the accuracy and scalability of algorithms can be improved. Though the clustering methods for collaborative filtering have plenty of applications, they have long been neglected. The similarity measure for clustering was confused with the similarity measure for collaborative filtering. In this paper, we propose a new similarity measure for clustering and its application. Firstly, we use a basic similarity function to discover neighbor vectors of items. Secondly, we calculate the cosine similarity of the neighbor vectors for clustering. Thirdly, we finish the clustering process by using adjusted DBSCAN. For those users having many known ratings, we adjust the prediction function by adding a parameter which is the function of the item cluster size. The experiments on the grouplens dataset show that our method outperforms the previous methods and has been proved in many clustering applications in the field of collaborative filtering. For those users having many known ratings, our application method of clustering improves the prediction accuracy.

A Requirement Analysis Pattern Selection Method for E-Business Project Situation

During the past decade, information systems investment has grown rapidly worldwide and information systems project development has become one of the most important targets in e-business. Yet, the failure of information systems projects is a common occurrence in many organizations around the world. A theoretical model is derived based upon organizational learning theory and prior research in order to examine the effects of organizational learning on lack of teams expertise risk. A survey method is applied to test the hypotheses proposed by the research model, and Taiwanese corporate companies serve as examples. After survey by questionnaire and analyze the data by structure equation modeling, the result reveals that organizational learning has significantly negative impacts on all of the lack of development expertise risk, lack of domain expertise risk, and lack of general expertise risk. These findings support information systems managers with valuable information to revisit their priorities in terms of the relative efforts in organization learning.Current situations of the e-business project are variety. One of the most difficult jobs for requirements engineers is to adopt the best appropriate RE analysis patterns for the e-business project at hand. It is possible to characterize typical e-business project situation which could be used as a base for selecting best appropriate patterns. To select best appropriate patterns, in this paper, the method was proposed including the pattern clustering analysis and the filtering recommendation for the project situation. Through the contrast with selection method based experience, the new method is feasible and effective.

Research on Testing-Based Software Credibility Measurement and Assessment

It’s one of the important approaches to measure and assess software credibility by testing in trustworthy software study. From the perspective of effective management and credibility analysis supporting the test process, this article describes a basic framework for its management and discusses the assessment techniques and methods of credible test process and software product.

A new modeling method of component interaction behavior

The behavioral semantics of component interfaces plays a significant role in the reuse, integration, validation and maintenance of components. However, the interface description specification of COTS component usually only contains syntax information, and lacks behavioral semantics information. This paper presents a method, called EKBehavior, for incrementally inferring a behavior model that represents the interplay between variable values and method invocations. EKBehavior extends the KBehavior algorithm for extracting finite state automata from execution traces, to the case of finite state automata with parameters. Popular dynamic analysis techniques synthesize either information on the values of the variables or information on relations between invocations of methods. Thus, these approaches do not capture the important relations that exist between data values and invocation sequences. Compared with the existing techniques, method generates model that represents the sequence constraints of invocations of methods, constraints of variable values, and the relations between variable values and method invocations.

Requirements engineering knowledge share and retrieve based on situations similarity

Project situation is one critical key foundation of requirements engineering (RE) knowledge share and reuse. RE knowledge can be shared and reused under similar project situation. A new knowledge retrieve technology based on project situation similarity is proposed in this paper. Firstly, the tree models or hierarchy models of RE knowledge situation are presented. Secondly, three situation match schemes are summarized and analyzed with reference to the tree matching, including tree inset, tree inclusion and tree intersection. Further, a quick tree matching algorithm is provided upon these three schemes. Through algorithm application, the results show that this solution is feasible and effective for the knowledge retrieve with high recall and precision.

Study and implementation of hybrid scheduling algorithm on JSP

This paper analyzes the requirement of a practical job-shop scheduling, which is different from the classical job-shop scheduling problems (JSP), more focus on strict resource constraints. First is formulated a mathematical model. Then described a new method, combined the time Petri net and the hybrid genetic algorithm, to solution of the practical JSP. The time Petri net was used to model discrete events in job shop scheduling, the scheduling results was obtained based on hybrid genetic algorithm. The objective of the new method is obtained makespan minimization in JSP. At the end of the paper, a practical example is described, and the results show that the proposed method is effective.

Modeling interaction constraints of component interfaces based on dynamic analysis

The interface specification of components usually lacks behavioral semantics information. Behavioral semantics information specify component interaction constraints on the usage of interface methods, including the order in which these methods should be called and the constraints on the method arguments and returns of methods. Based on dynamic analysis technique, a method which abstracts interaction constraints of component interfaces from the execution traces of component-based software system test is presented in this paper. These interaction constraints are incrementally modeled as finite state machines with parameters. Compared with the existing techniques which synthesize either information on the values of the variables or information on sequences of method invocations, our method generates models that represent the sequence constraints of method invocations, constraints of variable value, and the relations that exist between data values and method invocations.

Requirements analysis knowledge share and reuse on situation

One of the most difficult jobs for requirements analysis engineers is to select the appropriate requirements analysis knowledge for the project at hand. Once the selected requirement analysis knowledge does not fit the current project situation, it is doomed that some requirements problems will be emergent and the final system canpsilat fit in with the enterprise needs at a great extent. Requirements analysis knowledge can be shared and reused under similar project situation. A new knowledge reuse method based on project situation similarity is proposed in this paper. Firstly, the tree models or hierarchy models of requirements analysis knowledge situation are presented. Secondly, three situation match schemes are summarized and analyzed with reference to the tree matching, including tree inset, tree inclusion and tree intersection. Further, a quick tree matching algorithm is provided upon these three models. Through the method application, the results show that it is feasible and effective for the improvement of the knowledge reuse quality.