Abraham Ajith - Intelligent Data Analysis and Applications Proceedings of the Second Euro-China Conference on Intelligent Data Analysis and Applications, ECC 2015

Intelligent Data Analysis and Applications Proceedings of the Second Euro-China Conference on Intelligent Data Analysis and Applications, ECC 2015 — read online for free the complete book (whole text) full work

Below is the text of the book, divided by pages. System saving the place of the last page read, allows you to conveniently read the book "Intelligent Data Analysis and Applications Proceedings of the Second Euro-China Conference on Intelligent Data Analysis and Applications, ECC 2015" online for free, without having to search again every time where you left off. Put a bookmark, and you can go to the page where you finished reading at any time.

Light

Font size:

↓

↑

Georgia Tahoma Arial Verdana

Reset

Interval:

↓

↑

Bookmark:

Make

1234567...161

Part I
Data Analysis and Applications

Springer International Publishing Switzerland 2015

Ajith Abraham , Xin Hua Jiang , Vclav Snel and Jeng-Shyang Pan (eds.) Intelligent Data Analysis and Applications Advances in Intelligent Systems and Computing 10.1007/978-3-319-21206-7_1

Association-Rule-Based Random Walk Method for Personalized Tag Recommendation

Jing Wang 1

(1)

Information Technology Center, Tsinghua University, 100084 Beijing, China

Jing Wang (Corresponding author)

Email:

Nianlong Luo

Email:

Abstract

With the development of social websites during the web2.0, tagging has been playing an important role for users to mark their web resource. By offering personalized tags, recommender systems help users to integrate resource and complete tagging effectively. Graph based methods of tags recommending have been shown to provide high quality results such as RWR, FolkRank, PageRank, etc. However, data sparsity leads to sparseness of graphs limiting the precision in the process of use. In this paper, we propose a new method ARRW to alleviate the sparsity problem. We introduce Association Rules to Random Walk for digging up more relevance among nodes in graphs. We evaluate ARRW on a real-world dataset collected on Delicious. Data tests show that ARRW outperforms other Random Walk methods which not consider intra-relations, and ARRW successfully alleviate the graph sparsity problem.

Keywords

Association rule Tag recommendation Random walk Data sparsity

Introduction

Social tagging system is an application which provides a function of marking labels for web users. With the prevalence of web2.0, content-sharing websites of which social tagging mechanism is a key part develop rapidly. The mechanism transfers the right of categorizing internet resources from specialists to common persons. Users can freely provide metadata to describe the content of resource on the Internet so as to organize the resource easily and share with their friends conveniently. There are several typical websites with tagging systems, such as Delicious with publications.

Personalized tag recommendation [] is the main part in a tagging system. The systems predict and offer users a set of tags that they are most likely to use. Users tagging behaviors can be seen as ternary relations . When a Tag t is recommended for a User u to annotate Item i , t should be highly relevant to both u and i . For the reason that different users have diverse ways of tagging, recommended tags should differ from user to user. At the same time, the tags should be able to describe the content of items.

Our work builds on the personalized tag recommender method by using Random Walk model. Methods based on Random Walk like Random Walk with Restart (RWR) [] have shown to result in good prediction quality. But the sparsity problem still exists in graph, which limit the prediction quality. Here, we introduce association rules into Random Walk, in order to acquire more semantic information in the graph.

In this paper, our contributions are summarized as follows:

Building a new model (Association Rule based Rand Walk) for personalized tag recommendation.

Conducting experiments on the real-world dataset and demonstrating that our model outperforms other models based on Random Walk.

Our method can alleviate the problem of sparsity.

The rest of this paper is organized as follows. In the next section, we review and discuss the related work. Then, in Sect..

Related Work

Personalized tag recommendation is a hot topic in recommender system over the years. Different types of tagging systems were detailed by Huberman et al. []. In general, according to whether the processing procedure involves the resource content, tag recommendation methods fall into two categories: content-based approaches and graph-based approaches.

Content-based methods, which usually collect available information from context of items (e.g., web pages, anchor text, academic papers or other textual resources) to build user models or item models, can predict tags even for cold start. Yin [] proposed a Bayesian probabilistic model, the prediction is treated as the reverse of web search, consider a list of words on web pages as a list of tags, then retrieve the potential tags for the given web page.

Liu et al. [] put forward a fast tag recommending framework named Feature-Driven Tagging which represents a tag by some features. The feature can be a word, an id or other context information. Content-based approaches usually result in better precision than non-content-based approaches in most cases because of the more information ultilized by the former. However, content-based methods are unavailable where items belong to nonstructural resources or the content can not be obtained directly, such as movies, songs, etc.

Graph-based methods, which focus on the relations between users, items and tags, mostly yield lower computation complexity compared with content-based methods, for they dont parse items content. Adriana et al. [] extended Latent Dirichlet Allocation for recommending tags, which is on the hypothesis that item content consists of some latent topics. When a topic is represented by a tag probability distribution, tags can be predicted according to the posterior probability of latent topics.

There are many methods based on Factorization models. Higher-Order-Singular-Value-Decomposition (HOSVD) was introduced into tag recommendation to reduce tensor dimensionality in [] put forward a special Tucker Decomposition model called Pairwise Interaction Tensor Factorization (PITF) with linear runtime for learning and prediction. On their experiment dataset, the run time of PITF is shorter than TD, and the former can get better prediction as well. Since methods based on Factorization are quite complex, they are hardly applied to real systems.

In the fields of item recommendation like e-commerce, movie, music or photo websites, collaborative filtering [] introduced collaborative filtering into tag recommendation.

There were some models considering tag recommendation from other perspectives. Jin [].

Definition

When users are marking labels to items, tag recommender systems will predict and offer some tags the users might use. The systems calculate target tags from users behavior history and item context.

Given a set of items I , tags T , and users U , we represent the users marking to items with the ternary relations . For tag recommender, the task is to offer a specific pair (user,item) a list of tags. Here, we define distinct user-item pairs Ps : . All methods presented in this paper, give a scoring function

Next page

1234567...161

Light

Font size:

↓

↑

Georgia Tahoma Arial Verdana

Reset

Interval:

↓

↑

Bookmark:

Make

Similar books «Intelligent Data Analysis and Applications Proceedings of the Second Euro-China Conference on Intelligent Data Analysis and Applications, ECC 2015»

Look at similar books to Intelligent Data Analysis and Applications Proceedings of the Second Euro-China Conference on Intelligent Data Analysis and Applications, ECC 2015. We have selected literature similar in name and meaning in the hope of providing readers with more options to find new, interesting, not yet read works.

Fausto Pedro García Márquez

Computational Intelligence, Data Analytics and Applications: Selected papers from the International Conference on Computing

Pallavi Vijay Chavan

Data Science: Techniquest and Intelligent Applications

Garima Mathur (editor)

Proceedings of 2nd International Conference on Artificial Intelligence: Advances and Applications: ICAIAA 2021 (Algorithms for Intelligent Systems)

Nikhil Marriwala

Soft Computing for Intelligent Systems: Proceedings of ICSCIS 2020

Jyotsna Kumar Mandal

Proceedings of International Conference on Advanced Computing Applications: ICACA 2021

Soham Sarkar (editor)

Intelligent Multi-Modal Data Processing (The Wiley Series in Intelligent Signal and Data Processing)

Debahuti Mishra

Intelligent and Cloud Computing: Proceedings of ICICC 2019, Volume 1

De-Shuang Huang

Intelligent Computing Theories and Application: 17th International Conference, ICIC 2021, Shenzhen, China, August 12–15, 2021, Proceedings, Part II

Jianchao Zeng

Data Science: 7th International Conference of Pioneering Computer Scientists, Engineers and Educators, ICPCSEE 2021, Taiyuan, China, September 17–20, ... in Computer and Information Science, 1452)

Honghao Gao

Intelligent Mobile Service Computing

Abraham Ajith.

Intelligent Systems Design and Applications: 16th International Conference on Intelligent Systems Design and Applications (ISDA 2016) held in Porto, Portugal, December 16-18, 2016

Nilanjan Dey

Intelligent Speech Signal Processing

Discussion, reviews of the book Intelligent Data Analysis and Applications Proceedings of the Second Euro-China Conference on Intelligent Data Analysis and Applications, ECC 2015 and just readers' own opinions. Leave your comments, write what you think about the work, its meaning or the main characters. Specify what exactly you liked and what you didn't like, and why you think so.