Search

Applications like customer identification from their peculiar purchase patterns require class-wise discriminative feature subsets called as class signatures for classification. If the classifiers like KNN, SVM, etc. which require to work with a complete feature set, are applied to such applications, then the entire feature set may introduce errors in the classification. Decision tree classifier generates class-wise prominent feature subsets and hence, can be employed for such applications. However, all of these classifiers fail to model the relationship between features present in vector data. Thus, we propose to model the features and their interrelationships as graphs. Graphs occur naturally in protein molecules, chemical compounds, etc. for which several graph classifiers exist. However, multivariate data do not exhibit the graphs naturally. Thus, the proposed work focuses on (1) modeling multivariate data as graphs and (2) obtaining class-wise prominent subgraph signatures which are then used to train classifiers like SVM for decision making. The proposed method dSubSign can also classify multivariate data with missing values without performing imputation or case deletion. The performance analysis of both real-world and synthetic datasets shows that the accuracy of dSubSign is either higher or comparable to other existing methods.

The article deals with the issues of customer identification using new modern technologies that make it possible for banks to carry out more effective measures to counteract the legalization (laundering) of proceeds from crime. The author analyzes the current Russian legislation and regulations of the Bank of Russia in terms of remote identification of individual customers for individual banking operations. The mechanism of remote identification is considered, within which a technological infrastructure has been developed, including a Single biometric system and a Single identification and authentication system. The article notes that the introduction of remote identification provides advantages for both banks and Bank clients. The ability of banks to use identification data obtained by other credit institutions is the main goal of changes to the remote identification procedure. At the same time, credit organizations in such cases, taking into account the recommendation of the Basel Committee, should not forget to implement the “know your customer” policy. The paper examines the practice of Russian banks in the framework of pilot projects on the use of blockchain technology for remote identification and storage of information in relation to customers, in particular, on the example of a pilot KYC Project based on Masterchains. The author identifies the main legal risks of using new digital technologies, and emphasizes the need to ensure the safety of personal data.

Public distribution scheme is a structure promoted by government and includes chain of shops. This scheme aims to provide food grains and commodities to people at affordable price. Public distribution scheme ails at many factors such as availability, overcharge, timing, undersupply illegal smuggling of goods and corruption. Intelligent Public Ration Distribution System (IPRDS) is similar to an Automated Vending Machine which allows customers to complete purchases and transactions of basic ration commodities with ease. The customer identification, distribution of commodities, billing and data updation etc. are automated. The customer identification is done by the finger print verification. The system was prototyped and tested to distribute three major commodities; rice, wheat and kerosene. The proposed scheme can be directly implemented in the existing ration shops and thereby reduce the labor works associated with distribution of commodities. Since all the activities will be monitored by the food supplies department, all activities takes place in accordance with the rules and regulations.

Class imbalance brings great challenges to feature selection in customer identification and most of current feature selection approaches cannot produce good prediction on the minority class. A number of studies have attempted to solve this issue by using resampling techniques. However, resampling techniques only use the in-domain information and they cannot achieve good performance when the imbalance is caused by the absolute rarity of the minority class. In this paper, we focus on the issue of feature selection with class imbalance caused by absolute rarity. By introducing the idea of transfer learning, we develop a transferred feature selection method based on the group method of data handling neural network (GMDH). The proposed ensemble neural network extracts information of similar customers from related domains to deal with the information scarcity of the minority class in the target domain. Experiments are done on a real-world application from cigarette company. The results indicate that the new method gives better predictive performance than other benchmark feature selection methods, especially the predictive accuracy of minority high-value customers. At the same time, the new algorithm can help to identify important features that distinguish high-value customers from low-value ones. ispartof: Neural Computing & Applications vol:28 issue:9 pages:2593-2603 status: published

RFM analysis and market basket analysis (i.e., frequent pattern mining) are two most important tasks in database marketing. Based on customers’ historical purchasing behavior, RFM analysis can identify a valuable customer group, while market basket analysis can find interesting purchasing patterns. Previous studies reveal that recency, frequency and monetary (RFM) analysis and frequent pattern mining can be successfully integrated to discover valuable patterns, denoted as RFM-customer-patterns. However, since many retailers record transactions without collecting customer information, the RFM-customer-patterns cannot be discovered by existing approaches. Therefore, the aim of this study was to define the RFM-pattern and develop a novel algorithm to discover complete set of RFM-patterns that can approximate the set of RFM-customer-patterns without customer identification information. Instead of evaluating values of patterns from a customer point of view, this study directly measures pattern ratings by considering RFM features. An RFM-pattern is defined as a pattern that is not only occurs frequently, but involves a recent purchase and a higher percentage of revenue. This study also proposes a tree structure, called an RFM-pattern-tree, to compress and store entire transactional database, and develops a pattern growth-based algorithm, called RFMP-growth, to discover all the RFM-patterns in an RFM-pattern-tree. Experimental results show that the proposed approach is efficient and can effectively discover the greater part of RFM-customer-patterns.

Plain contextual approach does not always outperform un-contextual one.Comparison depends on effects of contextual variable on customers' behavior.Using different contexts for different users lead to best performance.We propose plain and "smart" contextual re-identification approaches.Framework for selecting the best performing contextual approach for re-identification. The task of re-identification consists of linking records for individuals with no identifying information to records with identifying information (i.e., name or social security number) in order to identify individuals within the anonymous data. This task is important for business since firms want to precisely identify consumers for several reasons, such as targeting advertisements to them or labeling them as fraudulent users. For these reasons, companies strive to improve their re-identification techniques. In addition, the re-identification task is relevant from a research prospective, and many algorithms and techniques have been proposed to improve existing re-identification models. However, no previous research has studied whether the use of contextual variables can improve re-identification performance. Context can be defined as the circumstances under which transactions take place. To date, contextual information (i.e., the time of day when or the location where digital data was created) has been used successfully in other modeling tasks such as in the recommender system domain, where its ability to improve the accuracy of lists of items suggested to website users has been demonstrated. Including contextual information in a re-identification model is not a trivial task for several reasons. In this paper, we discuss the main issues regarding the use of context for the re-identification task, namely, when incorporating context is expected to help re-identification and when it is expected to hurt. We propose contextual re-identification models and a framework for deciding when to use these and determining the best performing contextual method for the re-identification task. We test our contextual models using three different case studies. Our findings have a significant impact on expert and intelligent systems since they provide the first evidence of the possibility of including contextual variables for improving the results of the re-identification process. The results also have a relevant impact for businesses since they can help managers decide when and how to include a contextual variable into the re-identification task and contextualize subsequent actions after the re-identification task.

Classification is one of the most popular behavior prediction tools in behavior informatics (behavior computing) to predict group membership for data instances. It has been greatly used to support customer relationship management (CRM) such as customer identification, one-to-one marketing, fraud detection, and lifetime value analysis. Although previous studies showed themselves efficient and accurate in certain CRM classification applications, most of them took demographic, RFM-type, or activity attributes as classification criteria and seldom took temporal relationship among these attributes into account. To bridge this gap, this study takes customer temporal behavior data, called time-interval sequences, as classification criteria and develops a two-stage classification framework. In the first stage, time-interval sequential patterns are discovered from customer temporal databases. Then, a time-interval sequence classifier optimized by the particle swam optimization (PSO) algorithm is developed to achieve high classification accuracy in the second stage. The experiment results indicate the proposed time-interval sequence classification framework is efficient and accurate to predict the class label of new customer temporal data.

Despite the importance of data mining techniques to customer relationship management (CRM), there is a lack of a comprehensive literature review and a classification scheme for it. This is the first identifiable academic literature review of the application of data mining techniques to CRM. It provides an academic database of literature between the period of 2000-2006 covering 24 journals and proposes a classification scheme to classify the articles. Nine hundred articles were identified and reviewed for their direct relevance to applying data mining techniques to CRM. Eighty-seven articles were subsequently selected, reviewed and classified. Each of the 87 selected papers was categorized on four CRM dimensions (Customer Identification, Customer Attraction, Customer Retention and Customer Development) and seven data mining functions (Association, Classification, Clustering, Forecasting, Regression, Sequence Discovery and Visualization). Papers were further classified into nine sub-categories of CRM elements under different data mining techniques based on the major focus of each paper. The review and classification process was independently verified. Findings of this paper indicate that the research area of customer retention received most research attention. Of these, most are related to one-to-one marketing and loyalty programs respectively. On the other hand, classification and association models are the two commonly used models for data mining in CRM. Our analysis provides a roadmap to guide future research and facilitate knowledge accumulation and creation concerning the application of data mining techniques in CRM.

Personal social networks are considered as one of the most influential sources in shaping a customer's attitudes and behaviors. However, the interactions with friends or colleagues in social networks of individual customers are barely observable in most e-commerce companies. In this paper, we study the problem of customer identification in social networks, i.e., connecting customer accounts at e-commerce sites to the corresponding user accounts in online social networks such as Twitter. Identifying customers in social networks is a crucial prerequisite for many potential marketing applications. These applications, for example, include personalized product recommendation based on social correlations, discovering community of customers, and maximizing product adoption and profits over social networks. We introduce a methodology CSI (Customer-Social Identification) for identifying customers in online social networks effectively by using the basic information of customers, such as username and purchase history. It consists of two key phases. The first phase constructs the features across networks that can be used to compare the similarity between pairs of accounts across networks with different schema (e.g. an e-commerce company and an online social network). The second phase identifies the top-K maximum similar and stable matched pairs of accounts across partially aligned networks. Extensive experiments on real-world datasets show that our CSI model consistently outperforms other commonly-used baselines on customer identification.

The study of customer identification has the extremely vital significance to promote the core competitiveness of the enterprise. This paper focus on the problem of feature selection in customer identification. We try to solve the issue of feature selection under class imbalance and a hybrid method is proposed. We improve the data cleaning technology Tomek Links and get a new model called I-Tomlinks. Based on the using of I-Tomlinks for data preprocessing, we combine the group method of data handling (GMDH) and transfer learning together to construct a new feature selection model to solve the problem of class imbalance. The experiments show that the new method gives better predictive performance that other methods used as benchmarks. The new model provides a new tool for customer identification.

Applications of adaptive neural network models, i.e. models that are updated while they are used, are becoming more widespread. Adjustments to an adaptive model are based on a feedback which is collected while the model is in operation. While changing model parameters in production can improve the performance, itcomes with inherent risks. In particular, any erroneous adjustment to an online model may reduce the performanceand be detrimental to control applications and costly to the business. At the same time, mistakes in feedback are sometimes unavoidable. Therefore, an adaptively trained model with old data is sub-optimal unless the new revisions are taken into account. In this paper, we investigate the effect of a faulty feedback on the performance of an e-commerce customer identification neural network model. We first investigate the impact of feedback error on an adaptive model's performance. We then examine a technique to undo the incorrect adjustments to the model by re-training the adaptive model by a corrected feedback. Our results showthe majority of lossinmodel performance due to the feedback error is recoveredby re-training the adaptive model with the new corrected data.

This paper proposes a comprehensive classification framework applicable to the analytical Customer Relationship Management (CRM) problem domain of customer identification. Effective data mining tools have for long been anticipated in CRM as a promising technique to extract from historical data the knowledge that improves the quality of all CRM functions. However, standardized CRM data mining processes are yet to be developed. The proposed methodology provides quality solutions to most challenges encountered during a typical analytical CRM project, and has been tested on the difficult task from the UC San Diego Data Mining contest. The result outperforms some prevalent data mining techniques in the CRM domain.

This paper proposes the theoretical framework and model of decision making of customer retention based on customer identification. First, it proposes the 3-dimentional framework model of customer identification, which incorporates the dimensions of Customer Lifetime Value, Customer Perceived Value, and Accessibility under Competition. Second, it analyses the significance of customer identification to customer retention theoretically through establishing customer identification model. Third, it explains the procedure and method of customer retention decision-making based on customer identification. Finally, by establishing the decision-making model of customer retention, it analyses the pricing strategies incorporating the factor of customer loyalty under two pricing modes respectively, which are uniform pricing and discriminating pricing. The customer retention strategy basing on customer identification can improve the efficiency of customer retention greatly, so, it is of great instructive significance to the implementation of CRM.

Traditional customer relationship management (CRM) models often ignore the correlation that could exist in the purchasing behavior of neighboring customers. Instead of treating this correlation as nuisance in the error term, a generalized linear autologistic regression can be used to take these neighborhood effects into account and improve the predictive performance of a customer identification model for a Japanese automobile brand. In addition, this study shows that the level on which neighborhoods are composed has an important influence on the extra value that results from the incorporation of spatial autocorrelation.

Imbalanced data problem has a significant impact on the performance of RVM pattern recognition. Customer identification is an important application domain of pattern recognition which is mapping from samples to different categories by machine learning. In order to solve the problem, the paper proposes a method named up-sampling which overcomes the phenomenon that the machine is more partial to the majority classes while ignoring the sparse and decreases the false judgment about the sparse ones.

This paper explores the feasibility of using machined vision based cash register system in a university cafeteria and automatically determine price of the food items that are contained on a cafeteria food tray. The idea is to take an image of the food tray, identify all the food items on the tray, and display the total food cost. A fully automated version of the system is envisioned to use a standard biometric technology for customer identification and for automatic debit process of the food cost from the customer's pre-registered account. The analysis method uses shape and size analysis to identify food items that are served on distinct type dishes and color analysis for those items that are served in similar type dishes. The paper explains prototype of a developed system and experimental results for 10 types of meals served at Ritsumeikan Asia Pacific University Cafeteria to justify the validity of this approach.

Customer Relationship Management is one of the leading business strategies for today's companies, the key to successful implementation of CRM is Customer Intelligence. This paper designs and implements a customer intelligence system based on improving LTV model and data mining. Two data mining techniques are used including Self-Organizing Map and Fuzzy Decision Tree. The proposed system provides such functions as customer identification, customer loyalty analysis, customer satisfaction analysis, profitable customer segmentation, and customer differentiation. It can be used to make customer strategies.

While a reliable protection against illegal copies does not exists today, tracking of illegal copies and prove of ownership are important detection functions, which can be realized by using passive security mechanisms of digital watermarking. Recent research has identified many watermarking algorithms for all common media types ranging from printed matter to multimedia files. The main topics of interest concentrates on transparency and robustness: The watermark must not reduce media quality and should be detectable after most common media operations and attacks. Algorithm security is discussed with regards only to key space most of the times, while especially for customer identification known as active fingerprinting specialized attacks like coalition attacks are known. Digital fingerprinting raises the additional problem that we produce different copies for each customer. Attackers can compare several fingerprinted copies to find and destroy the embedded identification string by altering the data in those places where a difference was detected. Few approaches have been introduced for image and video watermarking schemes, but there are no observations for audio fingerprinting techniques. In our paper we discuss methods for secure customer identification by digital fingerprinting for audio data. We describe first two algorithms by Boneh et al. [BoSh95] and Schwenk et al. [DBS+99] and then combine these schemes with an audio watermarking algorithm for practical evaluation of their coalition resistance to detect illegal copies. We provide test results and evaluate the security against different types of coalition attacks.

In our publication we present a fingerprint watermark for the compressed video formats MPEG1/2 to ensure customer identification. The basic customer information is a watermark, which was developed for still images. We transfer it into the video area and adjust it to the special facts of MPEG video to be resistant against coalition attacks.