Abstract

We propose a supervised learning technique defining significant frequent patterns for associative regression. Assuming frequent patterns quantify correlations in dataset, we constrain the Generalized Iterative Scaling (GIS) convergence algorithm for Maximum Entropy (ME) models. We have used the combinations of ME parameters and GIS probabilities as discriminative weights to frequent patterns. The weighted frequent patterns then order the predictive analytics output. Experiments are conducted on sparse numeric datasets. Results suggest that condensed representations of frequent patterns allow parametric models suitable for class association rule mining.

Abstract

We present here, TagMiner, a data mining approach for part-of-speech (POS) tagging, an important Natural language process-ing (NLP) classification task. It is a semi-supervised associative clas-sification method for POS tagging. Existing methods for building POS taggers require extensive domain and linguistic knowledge and resources.Our method uses combination of a small POS tagged corpus and a raw untagged text data as training data to build the classifier model using association rules. Our tagger works well with very little training dataalso. The use of semi-supervised learning provides the advantage of notrequiring a large high quality tagged corpus. These properties make it es-pecially suitable for resource poor languages. Our experiments on various resource-rich, resource-moderate and resource-poor languages show good performance without using any language specific linguistic information.We note that inclusion of such features in our method may further im-prove the performance. Results also show that for smaller training data sizes our tagger performs better than state-of-the-art CRF tagger using same features as our tagger.

Abstract

We propose a supervised learning technique for associative regression. Assuming frequent patterns quantify correlations in dataset, we constrain the Sequential Conditional Generalized Iterative Scaling (SCGIS) convergence algorithm for Maximum Entropy (ME) models. We also assume prior probabilities on the ME model as a control on the step size in SCGIS. We have used the combinations of ME parameters and SCGIS probabilities as discriminative weights to frequent patterns. The weighted frequent patterns then predict the associative regression values. Experiments have been conducted on sparse numeric datasets, to find the regression error of our proposal. Our technique is comparable to the standard regression algorithms. As a concept, the proposed associative regression is useful as a parametric model in class association rule mining.

Abstract

Fuzzy Association Rule Mining (ARM) has been extensively used in relational or transactional datasets having less-to-medium number of attributes/dimensions. The mined fuzzy association rules (patterns) are not only used for manual analysis by domain experts, but are also leveraged to drive further mining tasks like classification and clustering which automate decision-making. Such fuzzy association rules can also be derived from high-dimensional numerical datasets, like image datasets, in order to train fuzzy associative classifiers or clustering algorithms. Traditional Fuzzy ARM algorithms are not able to mine rules from them efficiently, since such algorithms are meant to deal with datasets with relatively much less number of attributes/dimensions. Hence, in this paper we propose FAR-HD which is a Fuzzy ARM algorithm designed specifically for large high-dimensional datasets. FAR-HD processes fuzzy frequent itemsets in a DFS manner using a two-phased multiple-partition tidlist-based strategy. It also uses a byte-vector representation of tidlists, with the tidlists stored in the main memory in a compressed form (using a fast generic compression method). Additionally, FAR-HD uses Fuzzy Clustering to convert each numerical vector of the original input dataset to a fuzzy-cluster-based representation, which is ultimately used for the actual Fuzzy ARM process. FAR-HD has been compared experimentally with Fuzzy Apriori (7-15 times faster), which is the most popular Fuzzy ARM algorithm, and a Fuzzy ARM algorithm (1.1-4 times faster) which we proposed earlier and which is designed to work with very large but traditional (with fewer attributes) datasets.

Abstract

Regression is the study of functional dependency of one variable with respect to other variables. In this paperwe propose a novel regression algorithm, BINGR, for predicting dependent variable, having the advantage oflow computational complexity. The algorithm is interesting because instead of directly predicting the valueof the response variable, it recursively narrows down the range in which response variable lies. BINGRreduces the computation order to logarithmic which is much better than that of existing standard algorithms.As BINGR is parameterless, it can be employed by any naive user. Our experimental study shows that ourtechnique is as accurate as the state of the art, and faster by an order of magnitude

Abstract

Regression is the study of functional dependency of one nu-meric variable with respect to another. In this paper, we present a novel,efficient, binary search based regression algorithm having the advantageof low computational complexity. These desirable features make BINERa very attractive alternative to existing approaches. The algorithm isinteresting because instead of directly predicting the value of responsevariable, it recursively narrows down the range in which the responsevariable lies. Our empirical experiments with several real world datasetsshow that our algorithm, outperforms current state of art approachesand is faster by an order of magnitude.

Abstract

Measuring inter-document similarity is one of the most essential steps in text document clustering. Traditional methods rely on representing text documents using the simple Bag-of-Words (BOW) model. A document is an organized structure consisting of various text segments or passages. Such single term analysis of the text treats whole document as a single semantic unit and thus, ignores other semantic units like sentences, passages etc. In this paper, we attempt to take advantage of underlying subtopic structure of text documents and investigate whether clustering of text documents can be improved if text segments of two documents are utilized, while calculating similarity between them. We concentrate on examining effects of combining suggested inter-document similarities (based on inter-passage similarities) with traditional inter-document similarities following a simple approach for the same. Experimental results on standard data sets suggest improvement in clustering of text documents

Abstract

In this paper, we address the problem of active learning using the notion of in uence sets based on Reverse Nearest Neighbor. Active learning is an area of machine learning which emphasizes on achiev-ing optimal classi cation performance using as few labeled samples as possible. Reverse nearest neighbors have been used in domains such a sclustering and outliers detection in the past e ectively. In this paper, we devise a new sampling method for instances based on the knowledge pro-vided from RNN in uence sets. To demonstrate the eectiveness of our sampling method, we compare its performance against existing sampling methods on few real life datasets. The experimental results show that our technique outperforms existing methods, particularly on multi-class datasets

Abstract

We use text mining mechanisms to analyze Hot days in news feeds. We build upon the earlier work used to detect Hot topics and assume that we have already attained the Hot days. In this paper we identify the most relevant documents of a topic on a Hot day. We construct a similarity based technique for identifying and ranking these documents. Our aim is to automatically detect chains of hot correlated events over time.We develop a scheme using similarity measures like cosine similarity and KL-divergence to find correlation between these Hot days. For the ‘U.S. Presidential Elections’, the presidential debates which spanned over a week was one such event.

Abstract

Associative Classification leverages Association Rule Min-ing (ARM) to train Rule-based classifiers. The classifiers are built on high quality Association Rules mined from the given dataset. Associative Classifiers are very accurate be-cause Association Rules encapsulate all the dominant and statistically significant relationships between items in the dataset. They are also very robust as noise in the form of insignificant and low-frequency itemsets are eliminated dur-ing the mining and training stages. Moreover, the rules are easy-to-comprehend, thus making the classifier transparent.Conventional Associative Classification and Association Rule Mining (ARM) algorithms are inherently designed to work only with binary attributes, and expect any quantita-tive attributes to be converted to binary ones using ranges,like “Age = [25, 60]”. In order to mitigate this constraint,Fuzzy logic is used to convert quantitative attributes to fuzzy binary attributes, like “Age = Middle-aged”, so as to elimi-nate any loss of information arising due to sharp partition-ing, especially at partition boundaries, and then generate Fuzzy Association Rules using an appropriate Fuzzy ARM algorithm. These Fuzzy Association Rules can then be used to train a Fuzzy Associative Classifier. In this paper, we also show how Fuzzy Associative Classifiers so built can be used in a wide variety of domains and datasets, like transactional datasets and image datasets.