Haizhen Jiao’s research while affiliated with Beijing University of Posts and Telecommunications and other places

There are different types of social ties among people, and recognizing specialized types of relationship, such as family or friend, has important significance. It can be applied to personal credit, criminal investigation, anti-terrorism and many other business scenarios. So far, some machine learning algorithms have been used to establish social relationship inferencing models, such as Decision Tree, Support Vector Machine, Naive Bayesian and so on. Although these algorithms discover family members in some context, they still suffer from low accuracy, parameter sensitive, and weak robustness. In this work, we develop a Novel Family Relationship Recognition (NFRR) algorithm on telecom dataset for identifying one's family members from its contact list. In telecom dataset, all attributes are divided into three series, temporal, spatial and behavioral. First, we discover the most probable places of residence and workplace by statistical models, then we aggregate data and select the top-ranked contacts as the user's intimate contacts. Next, we establish Relational Spectrum Matrix (RSM) of each user and its intimate contacts to form communication feature. Then we search the user's nearest neighbors in labelled training set and generate its Specialized Family Spectrum (SFS). Finally, we decide family relationship by comparing the similarity between RSM of intimate contacts and the SFS. We conduct complete experiments to exhibit effectiveness of the proposed algorithm, and experimental results also show that it has a lower complexity.

For smartphone service providers, it is of vital importance to recognize characteristics of customers. The process of recognizing these characteristics is generally referred to as user profile, which provides knowledge basis for business decisions, enables intelligent services, and brings unique competitiveness. As a basic component of user profile, bedtime could reflect lifestyle, health condition, and occupation of people. This paper presents a flexible algorithm named BTP (Bedtime Prediction), which is designed for predicting wake time and bedtime by analysing screen status of smartphone. BTP first collects screen status log data of user’s smartphone and conducts preprocessing with a series of auxiliary user profiles. Then, it detects and records users’ wake time and bedtime of one day by searching and combining major screen extinguish periods in the past 24 hours. Finally, BTP predicts future bedtime by matching current screen status sequence with all historical records. By applying BTP, most of night and morning scenario-based applications could provide more considerate services, rather than following fixed execution time like alarm clock. Experiments on practical applications prove that BTP can effectively predict wake time and bedtime without applying complicated machine learning algorithms or uploading data to server.

Apriori is the most famous frequent pattern mining method. It scans dataset repeatedly and generate item sets by bottom-top approach. In order to reduce time complexity, we proposed a modified algorithm named as Frequent Matrix Apriori (FMA). Firstly, FMA scans the dataset only once to store frequent item information in a frequent matrix. Then, FMA discretize the matrix by the minimum support parameter which is generated automatically. Thirdly, it scans the discretized frequent matrix and find the most frequent item sets recursively. Experimental results proved that FMA is more effective than Apriori on time consuming with similar accuracy.

In many real world applications data is collected in multi-dimensional spaces, with the knowledge hidden in subspaces. It is an open research issue to select meaningful subspaces without any prior knowledge about such hidden patterns. Subspace clustering aims at detecting clusters in any projection of a high dimensional data space. However, almost all of the present subspace clustering methods cannot fnd subspace clusters with arbitrary shape, especially non-axis aligned clusters as we will demonstrate. In this work, we classify subspace clusters into three types: local dense clusters, axis-aligned clusters and nonaxis aligned clusters. To tackle the fundamental challenge of missing non-axis aligned clusters, we propose a new subspace clustering algorithm named SCUE (Subspace Clustering based on United Entropy). It computes each 1-dim entropy and united entropy of each two dimensions to form united entropy matrix. Cluster types are judged by entropy thresholds automatically generated from the matrix. Next it searches interesting subspaces in discretized united entropy matrix and gets clusters from interesting subspaces. Experimental results demonstrate that SCUE signifcantly outperforms present methods in both solution quality and effciency.

Real-time Fraud Detection has always been a challenging task, especially in financial, insurance, and telecom industries. There are mainly three methods, which are rule set, outlier detection and classification to solve the problem. But those methods have some drawbacks respectively. To overcome these limitations, we propose a new algorithm UAF (Usage Amount Forecast). Firstly, Manhattan distance is used to measure the similarity between fraudulent instances and normal ones. Secondly, UAF gives real-time score which detects the fraud early and reduces as much economic loss as possible. Experiments on various real-world datasets demonstrate the high potential of UAF for processing real-time data and predicting fraudulent users.