Search result diversification aims to offer diverse documents that cover as many intents possible. Most existing implicit approaches model diversity through the similarity of document representation, which is indirect and unnatural. To handle more precisely, we measure by their intent coverage. Specifically, build a classifier judge whether two different contain same based on document's content. Then construct an graph present complicated relationship query. On graph, are connected if they...

In recommender systems, measuring user similarity is essential for predicting a user's ratings on items. Most traditional works calculate the based historical shared between two users, without considering probability of users' different behaviors. To address this issue, our work designs measure behavior probabilities. Based method, complex network relationships modeled. The degree and community information modeled network, as well number are used with proposed to design rating prediction...

Link prediction provides methods to estimate potential connections in complex networks, which has theoretical and practical significance for personalized recommendation various other applications. Traditional collaborative filtering similar approaches have not utilized sufficient information on the community structure of networks. Therefore, this paper presents a link model based network modeling detection. In approach, networks are constructed by considering similarity among users'...

The K-shell has important theoretical significance and application value in measuring the importance of nodes complex networks. However, method, most possess an identical so that relative cannot be further compared with each other. Therefore, based on network values multi-order neighbors networks, this paper we use vectors to represent node which is named vector. Multi-order vector centrality defines a indicating number different K-shells groups them into elements Each infers not only...

Given that everyone online is saturated with information, the theoretical significance of recommendation algorithms evident in fact users need help finding products and content they care about. Collaborative filtering predicts a user's rating on an item by similar rated or items were user, using selected neighbors to "collaboratively filter" recommendation. In process, are considered equally important despite their differences popularity. Here, we explore method modeling recommender systems...

Search result diversification focuses on reducing redundancy and improving subtopic richness in the results for a given query. Most existing approaches measure document diversity mainly based text or pre-trained representations. However, some underlying relationships between query documents are difficult model to capture only from content. Given that knowledge base can offer well-defined entities explicit entities, we exploit relationship propose knowledge-enhanced search approach KEDIV....

This article studies the practical stability (PS) and finite-time (FTS) for fuzzy descriptor systems with uncertainties of unknown bound. For such nonlinear systems, novel sufficient conditions PS FTS are established. When follow proposed theorems, can be obtained alternatively. Meanwhile, linear matrix inequalities used, we devise adaptive controllers partially based on parallel-distributed compensation (PDC) non-PDC. Furthermore, numerical examples feasible region applied to inverted...

Stability analysis and stabilization for discrete-time singular delay systems are addressed, respectively. Firstly, a sufficient condition regularity, causality stability is derived. Then, by applying the skill of matrix theory, state feedback controller designed to guarantee closed-loop be regular, casual stable. Finally, numerical examples given demonstrate effectiveness proposed method.

The evolution of Internet topology is not always smooth but sometimes with unusual sudden changes. Consequently, identifying patterns critical for modeling and simulation. We analyze IPv6 in IP-level graph to demonstrate how it changes uncommon ways restructure the Internet. After evaluating average degree, path length, some other metrics over time, we find that case a large-scale growing becomes more robust; whereas top—bottom connection enhancement maintains its efficiency links largely decreased.

In this paper, finite-time robust control problems are investigated for Takagi– Sugeno fuzzy descriptor systems. We provide a novel sufficient condition admissibility of systems, involving group coupled linear matrix inequalities (LMIs), which guarantee dissipative performance, reduce conservativeness and improve LMIs forms. Second, corresponding controllers developed based on both parallel distributed compensation non-parallel compensation. Simulations demonstrate the effectiveness proposed...

The link prediction aims at predicting missing or future links in networks, which provides theoretical significance and extensive applications the related field. However, degree of confidence results has not been fully discussed works. In this article, we propose a similarity coefficient measure for prediction. former is used to balance reliability calculation results, might be untrustworthy due information asymmetry calculation, also makes it easier achieve optimal accuracy with smaller...

Slope One algorithm and its descendants measure user-score distance use the statistical score between users to predict unknown ratings, as opposed typical collaborative filtering that uses similarity for neighbor selection prediction. Compared systems select only similar neighbors, algorithms based on typically include all possible related in process, which needs more computation time requires memory. To improve scalability accuracy of distance-based recommendation algorithm, we provide a...

For nonlinear descriptor systems, this paper presents an approach to obtain a fuzzy controller with guaranteed finite-time stability and boundedness non-zero initial state, which outperforms some recent work additionally provides precision estimation of model approximation. We prove necessary sufficient conditions state for systems. Using Takagi–Sugeno dynamic models proposed conditions, we define sets use linear matrix inequalities satisfy differential inequalities. A simulation confirms...

Recommender systems are significantly useful to reduce information explosion nowadays. In order design an optimized algorithm, we develop improved opinion spreading approach predict online rating of recommender systems. The proposed method provides a solution zero-value problems similarity results, which is ignored in existing publications. could produce more precise prediction for each unrated user-item pair. our work, the items defined as number corresponding reviews user has given and...

Users’ ratings in recommender systems can be predicted by their historical data, item content, or preferences. In recent literature, scientists have used complex networks to model a user–user an item–item network of the RS. Also, community detection methods cluster users items improve prediction accuracy further. However, number links modeling is too large do proper clustering, and clustering NP-hard problem with high computation complexity. Thus, we combine fuzzy link importance K-core...