Data Mining Reading Group

Meeting 13 - Interestingness Measures for Data Mining: A Survey

2009-09-28T14:14:00.000-07:00

This week's paper is 'Interestingness measures for data mining: A survey' by Liqiang Geng and Howard Hamilton of University of Regina, Saskatchewan. This paper provides great deal of information about the notion of 'interestingness' and various different ways of quantifying it. Authors note that "Measuring the interestingness of discovered patterns is an active and important area of data mining research." and acknowledge that "..there is no widespread agreement on a formal definition of interestingness..." Their explanation of 'interest' is defined based on "definitions presented to-date" as a "broad concept that emphasizes conciseness, coverage, reliability, peculiarity, diversity, novelty, surprisingness, utility, and actionability. These nine specific criteria are used to determine whether or not a pattern is interesting."

We will discuss these areas and the Probability Based Objective Interestingness Measures mentioned in the paper such as Jaccard, Lift, Interestingness Weighting Dependency, Laplace Correction, Gini Index, Piatetsky-Shapiro, Cosine and Information Gain to name a few.

Paper's abstract is as follows.

Interestingness measures play an important role in data mining, regardless of the kind of patterns being mined. These measures are intended for selecting and ranking patterns according to their potential interest to the user. Good measures also allow the time and space costs of the mining process to be reduced. This survey reviews the interestingness measures for rules and summaries, classifies them from several perspectives, compares their properties, identifies their roles in the data mining process, gives strategies for selecting appropriate measures for applications, and identifies opportunities for future research in this area.

Meeting 12 - Improving Search In Social Networks by Agent Based Mining

2009-07-23T10:31:00.000-07:00

This week's paper is Anıl Gursel and Sandip Sen's "Improving Search In Social Networks by Agent Based Mining" from IJCAI 2009. The abstract of this paper is as follows.

"The popularity of social networks have burgeoned in recent years. Users share and access large volumes of information on social networking sites like Facebook, Flickr, del.icio.us, etc. Whereas a few of these sites have generic, impersonal searching mechanisms, we have developed an agent-based framework that mines the social network of a user to improve search results. Our Social Network based Item Search (SNIS) system uses agents that utilize the connections of a user in the social network to facilitate the search for items of interest. Our approach generates targeted search results that can improve the precision of the result returned from a user’s query. We have implemented the SNIS agent-based framework in Flickr, a photosharing social network, for searching for photos by using tag lists as search queries. We discuss the architecture
of SNIS, motivate the searching scheme used, and demonstrate the effectiveness of the SNIS
approach by presenting results. We also show how SNIS can be utilized for expertise location."

Meeting 11 - Topic Models - Latent Dirichlet Allocation

2009-05-11T16:35:00.000-07:00

One of the hot topics in the data mining world is how to discover the semantic meaning of a text document. Semantic meaning is related to moving beyond treating documents as a bag of words and instead discovering the underlying topics and meaning of the document.

One of the landmark papers is in this area is Latent Dirichelet Allocation by Jordan, Blei, and Ng, which introduces topic models for text classification.

Meeting 10 - Fast Mining of Distance-Based Outliers in High-Dimensional Datasets

2009-04-06T17:10:00.000-07:00

This week, Adnan will be presenting "Fast Mining of Distance-Based Outliers in High-Dimensional Datasets" by Amol Ghoting, Srinivasan Parthasarathy and Matthew Eric Otey of IBM, Ohio State University and Google Inc respectively.

Abstract: Defining outliers by their distance to neighboring data points has been shown to be an effective non-parametric approach to outlier detection. Existing algorithms for mining distance-based outliers do not scale to large, high-dimensional data sets. In this paper, we present RBRP, a fast algorithm for mining distance-based outliers, particularly targeted at high-dimensional data sets. RBRP scales log-linearly as a function of the number of data points and linearly as a function of the number of dimensions. Our empirical evaluation demonstrates that we outperform the state-of-the-art, often by an order of magnitude.

Keywords: Outlier detection, high-dimensional data sets, approximate k-nearest neighbors, clustering.

Meeting 9 - Support Vector Machines for Pattern Recognition

2009-03-13T12:54:00.000-07:00

Next week, Raja Peer is presenting the following paper on Support Vector Machines for the reading group.

A tutorial on support vector machines for pattern recognition
CJC Burges - Data mining and knowledge discovery, 1998 - Springer
http://www.cmlab.csie.ntu.edu.tw/~cyy/learning/papers/SVM_Tutorial.pdf

A very interesting topic, looking forward to it.

Meeting 8 - AdaBoost and Logistic Regression

2009-03-13T11:05:00.000-07:00

This week Jeff Bergman discussed the Adaptive Boosting and it's role as a technique for combining a number of “weak” classifiers to make a “strong” classifier.

The paper discussed was

A Short Introduction to Boosting
By Yoav Freund Robert E. Schapire
AT&T Labs Research
www.site.uottawa.ca/~stan/csi5387/boost-tut-ppr.pdf

and the following presentation was used as reference.

http://cmp.felk.cvut.cz/~sochmj1/adaboost_talk.pdf

Also, discussed was the successful application of AdaBoost to problem of Face Detection
in the classic paper, Rapid Object Detection using a Boosted Cascade of Simple Features.

Meeting 7 - Algorithms for Mining Distance-Based Outliers in Large Datasets

2009-02-26T13:12:00.000-08:00

This week our discussion is about the VLDB'98 conference paper on "Algorithms for Mining Distance-Based Outliers in Large Datasets." by Edwin M. Knox and Raymond T. Ng of University of British Columbia.

Paper Abstract is as follows.

This paper deals with finding outliers (exceptions) in large, multidimensional datasets. The identification of outliers can lead to the discovery of truly unexpected knowledge in areas such as electronic commerce, credit card fraud, and even the analysis of performance statistics of professional athletes. Existing methods that we have seen for finding outliers in large datasets can only deal efficiently with two dimensions/attributes of a dataset. Here, we study the notion of DB- (Distance- Based) outliers. While we provide formal and empirical evidence showing the usefulness of DB-outliers, we focus on the development of algorithms for computing such outliers. ( Proceedings of the 24th VLDB Conference)

Paper can be downloaded from VLDB conference website here.

Meeting 6 - Privacy Preserving Mining of Association Rules

2009-02-19T18:18:00.000-08:00

Paper:

Privacy Preserving Mining of Association Rules
By
Alexandre Evfimievski
Ramakrishnan Srikant
Rakesh Agrawal
Johannes Gehrke

In this meeting we went through the paper to understand mainly 2 randomization techniques (Select-a-size and cut-and-paste) and discussed the chanllenges of combining privacy preservation within data mining algorithms.

Lot of information to digest in a single paper.

Meeting 5 - Probabilistic Inference - 2/5/2009

2009-01-30T15:09:00.000-08:00

In the fifth meeting we plan to review probabilistic inference and modeling, focusing on Bayesian methods, including Bayesian Inference, Bayesian Networks, and Markov Random Fields, time permitting.

These concepts are fundamental for understanding various Data Mining techniques. Jeff Bergman will be presenting and reviewing the paper, Techincal Introduction: A Primer on Probabilistic Inference.

An additional paper that is a good overview is David Heckerman's, A Tutorial on Learning With Bayesian Networks.

Meeting 4 - Top Ten Data Mining Algorithms - 1/29/2009

2009-01-29T22:08:00.000-08:00

In the fourth meeting of Data mining reading group, we reviewed the top ten data mining algorithms paper based on the IEEE survey. Adnan Masood and Jeff Bergman co-presented the ten algorithms

Top 10 algorithms in data mining

Abstract: This paper presents the top 10 data mining algorithms identified by the IEEE
International Conference on Data Mining (ICDM) in December 2006: C4.5, k-Means, SVM,
Apriori, EM, PageRank, AdaBoost, kNN, Naive Bayes, and CART. These top 10 algorithms
are among the most influential data mining algorithms in the research community.

Meeting 3 - Map Reduce - 1/22/2009

2009-01-29T21:33:00.000-08:00

In the third meeting of Data Mining Reading Group, we reviewed the famous MapReduce paper from ACM SIGMod 2007. It was presented by Raja Peer.

Map-reduce-merge: simplified relational data processing on large clusters
International Conference on Management of Data archive
Proceedings of the 2007 ACM SIGMOD international conference on Management of data

Meeting 2 - Netflix Challenge & Collaborative Filtering - 1/15/2009

2009-01-29T21:28:00.000-08:00

In the 2nd data mining reading group meeting, we went over the Netflix Paper from KDD 2008. This week's presenter was Jeff Bergman.

Factorization Meets the Neighborhood: a Multifaceted Collaborative Filtering Model
by Yehuda Koren
AT&T Labs – Research
KDD’08, August 24–27, 2008, Las Vegas

Meeting 1 - Page Rank - 1/8/2009

2009-01-29T21:11:00.000-08:00

In the first data mining reading group meeting, we went over the Page Rank Paper. This week's presenter was Adnan Masood

The anatomy of a large-scale hypertextual Web search engine
S Brin, L Page - Computer Networks and ISDN Systems, 1998 - Elsevier

The pagerank citation ranking: Bringing order to the web
L Page, S Brin, R Motwani, T Winograd - 1998 - cpe.ku.ac.th
The PageRank Citation Ranking: Bringing Order to the Web.