4,510 Hits in 5.2 sec

Training Multiple Support Vector Machines for Personalized Web Content Filters

2013 IEICE transactions on information and systems  
We therefore propose two different strategies to train multiple SVMs for personalized Web content filters.  ...  The abundance of information published on the Internet makes filtering of hazardous Web pages a difficult yet important task.  ...  Acknowledgment We would like to thank the reviewers and the Associate Editor for their valuable comments and suggestions on the submission version of this paper.  ... 
doi:10.1587/transinf.e96.d.2376 fatcat:evabt2kalrbvxhrqowbjhvdjqm

News Retrieval through a MultiAgent System

Andrea Addis, Giuliano Armano, Francesco Mascia, Eloisa Vargiu
2007 Workshop From Objects to Agents  
The system is built upon a generic multiagent architecture that supports the implementation of personalized, adaptive and cooperative multiagent systems devised to retrieve, filter and reorganize information  ...  The continuous growth of information sources on the web, together with the corresponding volume of dailyupdated contents, makes the problem of finding news and articles a challenging task.  ...  ACKNOWLEDGMENTS This work has been supported by the Italian Ministry of Education, under the project "DART -Distributed Architecture for Semantic Search and Personalized Content Retrieval".  ... 
dblp:conf/woa/AddisAMV07 fatcat:in43e2gj6nazffwxpriegcczhy

Discriminative factored prior models for personalized content-based recommendation

Lanbo Zhang, Yi Zhang
2010 Proceedings of the 19th ACM international conference on Information and knowledge management - CIKM '10  
The standard Bayesian hierarchical model used in filtering assumes all user profiles are generated from the same Gaussian prior.  ...  The Bayesian hierarchical models learn user profiles jointly and have the advantage of being able to borrow information from other users through a Bayesian prior.  ...  To better model the diversity and commonality of users and each user's multiple interests, this paper proposes a flexible Bayesian hierarchical modeling approach for personalized content-based recommendation  ... 
doi:10.1145/1871437.1871674 dblp:conf/cikm/ZhangZ10 fatcat:bjmuhjiadfabpejivgihqclk74

Pairwise Webpage Coreference Classification Using Distant Supervision

S. Subramanian, Timothy Baldwin, Julian Brooke, Trevor Cohn
2017 Proceedings of the 26th International Conference on World Wide Web Companion - WWW '17 Companion  
A person or other entity is often associated with multiple URL endpoints on the web, motivating the task of determining whether a given pair of webpages is coreferent to a given entity.  ...  To strike a balance between unsupervised and supervised methods that require annotated data, we build a positive and unlabelled (PU) learning model, where we obtain positive examples using web search-based  ...  Additionally, we considered end-point pages only (filtered using features from [1] ), and use a random 70/30 split for training and testing.  ... 
doi:10.1145/3041021.3054224 dblp:conf/www/SubramanianBBC17 fatcat:tr4bxqamznelliaunywsvc4yiq

An Efficient Concept-based Mining Model for Deriving User Profiles

P. Sasikala, V. Vidhya
2012 International Journal of Applied Information Systems  
User profiling forms the basis for search engine personalization applications. Search engines are personalized so that they optimize the retrieval quality of user queries.  ...  Collaborative filtering filters information about a user based on a collection of user profiles that are already built from the extracted preferences.  ...  ,Feature_ c n ] for the ranking SVM training is composed of all the extracted concepts for a query q. For each concept c i , a feature vector is created ф(q,c i )=[Feature_c 1 ,Feature_c 2 ,...  ... 
doi:10.5120/ijais12-450187 fatcat:xifhtifycnb2rjxnij3ixe367u

A Method based on One-class SVM for News Recommendation

Limeng Cui, Yong Shi
2014 Procedia Computer Science  
In order to provide intelligent recommendation and personalized service for users on news website, this paper presents a method based on One-Class SVM for news recommendation algorithm.  ...  First, this algorithm preprocesses the webpages from Sogou Labs, each of which has its inherent domain and builds One-Class SVM models for these domains.  ...  According to domestic and international related literature, we have summarized the methods of recommendation system and personalized recommendation, such as content-based filtering and collaborative filtering  ... 
doi:10.1016/j.procs.2014.05.270 fatcat:rzh6lywdw5f55jczafnybnqetq

YouTubeCat: Learning to categorize wild web videos

Zheshen Wang, Ming Zhao, Yang Song, Sanjiv Kumar, Baoxin Li
2010 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition  
Extensive experiments on about 80K videos from 29 most frequent categories in YouTube show the effectiveness of the proposed method for categorizing large-scale wild Web videos 1 .  ...  A key issue is how to build an effective training set in the presence of missing, sparse or noisy labels.  ...  Multiple data sources As mentioned earlier, lack of labeled training data is a main bottleneck for general Web video categorization.  ... 
doi:10.1109/cvpr.2010.5540125 dblp:conf/cvpr/WangZSKL10 fatcat:e6ieenc53nhcliadzps6z4p3eu

Integrating Concept Ontology and Multitask Learning to Achieve More Effective Classifier Training for Multilevel Image Annotation

Jianping Fan, Yuli Gao, Hangzai Luo
2008 IEEE Transactions on Image Processing  
a multiple kernel learning algorithm is developed for SVM image classifier training.  ...  To tackle the problem of huge intraconcept visual diversity for the image concepts at the higher levels of the concept ontology, a novel hierarchical boosting algorithm is developed to learn their ensemble  ...  Schonfeld for handling the review process of this paper.  ... 
doi:10.1109/tip.2008.916999 pmid:18270128 fatcat:bxokhjalpzcfdpwyk3yajzl3ni

Refined experts

Paul N. Bennett, Nam Nguyen
2009 Proceedings of the 32nd international ACM SIGIR conference on Research and development in information retrieval - SIGIR '09  
While large-scale taxonomies -especially for web pageshave been in existence for some time, approaches to automatically classify documents into these taxonomies have met with limited success compared to  ...  Finally, we present an empirical study demonstrating that the suggested changes lead to 10-30% improvements in F1 scores versus an accepted competitive baseline, hierarchical SVMs.  ...  This combined with the fast-growing pace of the web as well as dynamically generated web-pages argues for the need for hierarchical classification methods that can automatically place web pages into a  ... 
doi:10.1145/1571941.1571946 dblp:conf/sigir/BennettN09 fatcat:ivek6ymadndknl6qn77jc3ogca

Evaluating tag filtering techniques for web resource classification in folksonomies

Nicolás Tourné, Daniela Godoy
2012 Expert systems with applications  
Furthermore, the use of several filtering and pre-processing operations to re- * Corresponding author  ...  , alleviating the task of manual classification commonly required by systems such as directories on the Web.  ...  (Vatturi et al., 2008 ) create a personalized tag-based recommender for each user consisting of two naïve Bayes classifiers trained over different time frames.  ... 
doi:10.1016/j.eswa.2012.02.088 fatcat:65rljaap5vdttd4clegggsyl3i

On the Role of Social Tags in Filtering Interesting Resources from Folksonomies

Daniela Godoy
2010 Lernen, Wissen, Daten, Analysen  
The results of using social tags for personal classification are compared with those achieved with traditional information sources about the user interests such as the textual content of Web documents.  ...  In this paper the problem of filtering resources from social tagging systems according to individual user interests using purely tagging data is studied.  ...  Acknowledgments This research was supported by The National Council of Scientific and Technological Research (CONICET) under grant PIP Nº 114-200901-00381.  ... 
dblp:conf/lwa/Godoy10 fatcat:k4j4d3465fdhjpmoof4ghkd3we

Deep classification in large-scale text hierarchies

Gui-Rong Xue, Dikan Xing, Qiang Yang, Yong Yu
2008 Proceedings of the 31st annual international ACM SIGIR conference on Research and development in information retrieval - SIGIR '08  
As a result, the classification model is trained on the small subset before being applied to assign the category for a new document.  ...  Experimental results show that our proposed approach can reach 51.8% on the measure of Mi-F1 at the 5th level, which is 77.7% improvement over top-down based SVM classification algorithms.  ...  INTRODUCTION Text classification is at the heart of Web page classification, which can find many applications ranging from Web personalization to targeted advertisements [1] on Web pages.  ... 
doi:10.1145/1390334.1390440 dblp:conf/sigir/XueXYY08 fatcat:p4edkcpa5zddhjoajl2e2yqm6y

Region-based automatic web image selection

Keiji Yanai, Kobus Barnard
2010 Proceedings of the international conference on Multimedia information retrieval - MIR '10  
Several works on Web image filtering task with bag-of-features have been proposed so far. However, in case that the training data includes much noise, sufficient results could not be obtained.  ...  In the experiments, we used a multiple-instance learning SVM and a standard SVM as discriminative methods, and pLSA and LDA mixture models as probabilistic generative methods.  ...  Assuming that a returned image set includes at least one positive image, they applied a multiple instance learning method for Web image filtering.  ... 
doi:10.1145/1743384.1743436 dblp:conf/mir/YanaiB10 fatcat:4etnxhwyejdcjffprkvmtkmsgy

Classification of Spam Emails through Hierarchical Clustering and Supervised Learning [article]

Francisco Jáñez-Martino, Eduardo Fidalgo, Santiago González-Martínez, Javier Velasco-Mata
2020 arXiv   pre-print
Finally, we recommend for the task of multi-class spam classification the use of (i) TF-IDF combined with SVM for the best micro F1 score performance, 95.39%, and (ii) TD-IDF along with NB for the fastest  ...  First, we applied a hierarchical clustering algorithm to create SPEMC-11K (SPam EMail Classification), the first multi-class dataset, which contains three types of spam emails: Health and Technology, Personal  ...  Regarding classifier parameters, we activated the class-weight and set a C value of 1000 for LR. For SVM, We selected "linear" as a kernel, C parameter to 1000, and activated the class-weight flag.  ... 
arXiv:2005.08773v2 fatcat:63i3qdbrhreznjpynzhl4hcyqq

An adaptive skin model and its application to objectionable image filtering

Qiang Zhu, Ching-Tung Wu, Kwang-Ting Cheng, Yi-Leh Wu
2004 Proceedings of the 12th annual ACM international conference on Multimedia - MULTIMEDIA '04  
We then use a Support Vector Machine (SVM) classifier to identify the skin Gaussian from the trained GMM by incorporating spatial and shape information of the skin pixels.  ...  Moreover, we examine how the improvement on skin detection by this adaptive skin-model impacts the detection accuracy in the application of Objectionable Image Filtering.  ...  Finally, we propose a novel idea of hierarchical bagging by using multiple classifiers to further improve the accuracy of objectionable image classifier.  ... 
doi:10.1145/1027527.1027538 dblp:conf/mm/ZhuWCW04 fatcat:hf6tgql5irdf3efdwf7xyese2e
« Previous Showing results 1 — 15 out of 4,510 results