Employing Structural and Textual Feature Extraction for Semistructured Document Classification

Khabbaz, Mohammad; Kianmehr, Keivan; Alhajj, R.

doi:10.1109/tsmcc.2012.2208102

Cited by 11 publications

(7 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Besides, it is a challenge to find straight forward parallel implementations of these optimization algorithms. In web-based scenarios, data changes very frequently [24]. This requires either algorithms with highly scalable training phase, or models we can sequentially update.…”

Section: Figure 1: Classification Example With 3 Classesmentioning

confidence: 99%

“…Other than problems we can directly model with convex hull, there are many other domains such as web mining where we deal with large graphs. We can also use properties of these domains such as link structure in order to define entities such as web pages in a multidimensional Euclidean space, and use convex hull for modeling [24].…”

Section: Convex Hull Backgroundmentioning

confidence: 99%

“…Data mining topics such as classification [1,2,14,15,19,24,25,29], clustering [38,28,27,26,24,8,7,5,3], frequent pattern mining [37,36,11], frequent sub-structure mining [11,24], regression [21,20], data cleaning [18,8,1], ranking [6], data warehousing [5], recommender systems [23,16,10], bio-informatics [19], outlier detection [22,13,4], nearest neighbors [17,9] and social networks [31,32], have been widely discussed in data management and prediction.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

DataGrinder: Fast, Accurate, Fully non-Parametric Classification Approach Using 2D Convex Hulls

Khabbaz¹

2015

Preprint

Self Cite

View full text Add to dashboard Cite

It has been a long time, since data mining technologies have made their ways to the field of data management. Classification is one of the most important data mining tasks for label prediction, categorization of objects into groups, advertisement and data management. In this paper, we focus on the standard classification problem which is predicting unknown labels in Euclidean space. Most efforts in Machine Learning communities are devoted to methods that use probabilistic algorithms which are heavy on Calculus and Linear Algebra. Most of these techniques have scalability issues for big data, and are hardly parallelizable if they are to maintain their high accuracies in their standard form. Sampling is a new direction for improving scalability, using many small parallel classifiers. In this paper, rather than conventional sampling methods, we focus on a discrete classification algorithm with O(n) expected running time. Our approach performs a similar task as sampling methods. However, we use column-wise sampling of data, rather than the row-wise sampling used in the literature. In either case, our algorithm is completely deterministic. Our algorithm, proposes a way of combining 2D convex hulls in order to achieve high classification accuracy as well as scalability in the same time. First, we thoroughly describe and prove our O(n) algorithm for finding the convex hull of a point set in 2D. Then, we show with experiments our classifier model built based on this idea is very competitive compared with existing sophisticated classification algorithms included in commercial statistical applications such as MATLAB.

show abstract

Section: Figure 1: Classification Example With 3 Classesmentioning

confidence: 99%

Section: Convex Hull Backgroundmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

DataGrinder: Fast, Accurate, Fully non-Parametric Classification Approach Using 2D Convex Hulls

Khabbaz¹

2015

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…SVM is the main classifier algorithm for text classification. It has performed more accurately than KNN, Naïve Bayes and Decision Tree in many applications [9][10][11]. It is widely used in web page classification and bioinformatics applications.…”

Section: Introductionmentioning

confidence: 99%

A technical study on Information Retrieval using web mining techniques

Sathiaseelan¹

2015

2015 International Conference on Innovations in Information, Embedded and Communication Systems (ICIIECS)

View full text Add to dashboard Cite

Information Retrieval (IR) is a significant area in web mining where the users procure their needed information from the web. The documents in the web are in the form of web pages. As there is a high usage of data in the web, researchers are intended to do research in the field of IR. Web query classification and web page classification are leading major role to classify the web documents in IR using classification algorithms, based on the user query. The classified documents are indexed using ranking algorithm. This research paper aims to study about various methods of IR to classify and index the relevant web pages respectively.

show abstract

“…2 Background techniques 2.1 Feature analysis in text classification The process of feature analysis acting as a decisive role in text classification involves the evaluation of document and the identification of meaningful properties in order to capture the most important patterns while eliminating irrelevant noise from text (Mladenic, 1998;Do et al, 2006;Saeys et al, 2007;Khabbaz et al, 2012). Conventionally, the method of evaluating the importance of textual aspects mainly concentrates on the term-level characteristics within the document.…”

Section: Introductionmentioning

confidence: 99%

A feature selection approach for automatic e-book classification based on discourse segmentation

Guo

Wang

Lai

2015

Program

View full text Add to dashboard Cite

Purpose -The purpose of this paper is to develop a novel feature selection approach for automatic text classification of large digital documentse-books of online library system. The main idea mainly aims on automatically identifying the discourse features in order to improving the feature selection process rather than focussing on the size of the corpus. Design/methodology/approach -The proposed framework intends to automatically identify the discourse segments within e-books and capture proper discourse subtopics that are cohesively expressed in discourse segments and treating these subtopics as informative and prominent features. The selected set of features is then used to train and perform the e-book classification task based on the support vector machine technique. Findings -The evaluation of the proposed framework shows that identifying discourse segments and capturing subtopic features leads to better performance, in comparison with two conventional feature selection techniques: TFIDF and mutual information. It also demonstrates that discourse features play important roles among textual features, especially for large documents such as e-books. Research limitations/implications -Automatically extracted subtopic features cannot be directly entered into FS process but requires control of the threshold. Practical implications -The proposed technique has demonstrated the promised application of using discourse analysis to enhance the classification of large digital documentse-books as against to conventional techniques. Originality/value -A new FS technique is proposed which can inspect the narrative structure of large documents and it is new to the text classification domain. The other contribution is that it inspires the consideration of discourse information in future text analysis, by providing more evidences through evaluation of the results. The proposed system can be integrated into other library management systems.2 PROG 49,1 digitalized format, which provides several benefits to readers, including shortened publication cycles, faster distribution channels that permit the wider propagation of timely information, and friendly visualization to display deliberate content of articles. As regarding the essence of content, e-book is different from other type of textual materials because it usually contains lengthy content which easily leads to higher feature dimension in the perspective of term-level analysis. As such, the book may consist of a variety of themes in the text stream, which are distributed and shifted among sentences or paragraphs in turns of the hidden subtopic to stretch the expression of the main topic (Hearst, 1997;Ridel and Bieman, 2012). Such phenomena mostly addressed in discourse analysis raises interesting issues to text processing problems for which an innovative strategy is needed as exploring the high-level linguistic attributes from e-book.The properties sustaining the lexical coherence of topic imply the flow of discourse information, which is valuable to be further investigated an...

show abstract

Employing Structural and Textual Feature Extraction for Semistructured Document Classification

Cited by 11 publications

References 28 publications

DataGrinder: Fast, Accurate, Fully non-Parametric Classification Approach Using 2D Convex Hulls

DataGrinder: Fast, Accurate, Fully non-Parametric Classification Approach Using 2D Convex Hulls

A technical study on Information Retrieval using web mining techniques

A feature selection approach for automatic e-book classification based on discourse segmentation

Contact Info

Product

Resources

About