• Journal of the Korean Society for information Management
• /
• v.26 no.3
• /
• pp.261-278
• /
• 2009

Identifying optimal feature sets in Text Categorization(TC) is crucial in terms of improving the effectiveness. In this study, experiments on feature expansion were conducted using author provided keyword sets and article titles from typical scientific journal articles. The tool used for expanding feature sets is WordNet, a lexical database for English words. Given a data set and a lexical tool, this study presented that feature expansion with synonymous relationship was significantly effective on improving the results of TC. The experiment results pointed out that when expanding feature sets with synonyms using on classifier names, the effectiveness of TC was considerably improved regardless of word sense disambiguation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.6 no.11
• /
• pp.2866-2879
• /
• 2012

Among the various security threats in online games, the use of game bots is the most serious problem. Previous studies on game bot detection have proposed many methods to find out discriminable behaviors of bots from humans based on the fact that a bot's playing pattern is different from that of a human. In this paper, we look at the chatting data that reflects gamers' communication patterns and propose a communication pattern analysis framework for online game bot detection. In massive multi-user online role playing games (MMORPGs), game bots use chatting message in a different way from normal users. We derive four features; a network feature, a descriptive feature, a diversity feature and a text feature. To measure the diversity of communication patterns, we propose lightly summarized indices, which are computationally inexpensive and intuitive. For text features, we derive lexical, syntactic and semantic features from chatting contents using text mining techniques. To build the learning model for game bot detection, we test and compare three classification models: the random forest, logistic regression and lazy learning. We apply the proposed framework to AION operated by NCsoft, a leading online game company in Korea. As a result of our experiments, we found that the random forest outperforms the logistic regression and lazy learning. The model that employs the entire feature sets gives the highest performance with a precision value of 0.893 and a recall value of 0.965.

• East Asian mathematical journal
• /
• v.37 no.5
• /
• pp.599-612
• /
• 2021

In China, Weibo is one of the social platforms with more users. It has the characteristics of fast information transmission and wide coverage. People can comment on a certain event on Weibo to express their emotions and attitudes. Judging the emotional tendency of users' comments is not only beneficial to the monitoring of the management department, but also has very high application value for rumor suppression, public opinion guidance, and marketing. This paper proposes a two-input Adaboost model based on TextCNN and BiLSTM. Use the TextCNN model that can perform local feature extraction and the BiLSTM model that can perform global feature extraction to process comment data in parallel. Finally, the classification results of the two models are fused through the improved Adaboost algorithm to improve the accuracy of text classification.

• Journal of Computing Science and Engineering
• /
• v.10 no.4
• /
• pp.103-110
• /
• 2016

Emotion recognition is an important component of affective computing, and is significant in the implementation of natural and friendly human-computer interaction. An effective approach to recognizing emotion from text is based on a machine learning technique, which deals with emotion recognition as a classification problem. However, in emotion recognition, the texts involved are usually very short, leaving a very large, sparse feature space, which decreases the performance of emotion classification. This paper proposes to resolve the problem of feature sparseness, and largely improve the emotion recognition performance from short texts by doing the following: representing short texts with word cluster features, offering a novel word clustering algorithm, and using a new feature weighting scheme. Emotion classification experiments were performed with different features and weighting schemes on a publicly available dataset. The experimental results suggest that the word cluster features and the proposed weighting scheme can partly resolve problems with feature sparseness and emotion recognition performance.

• Proceedings of the KIEE Conference
• /
• 2002.11c
• /
• pp.605-608
• /
• 2002

A text-dependent speaker recognition system using a robust matching process is studied. The feature histogram of LPC cepstral coefficients for matching is used. The matching process uses mixture network with penalty scores. Using probability and shape comparison of two feature histograms, similarity values are obtained. The experiment results will be shown to show the effectiveness of the proposed algorithm.

• Proceedings of the IEEK Conference
• /
• 2003.07e
• /
• pp.1835-1838
• /
• 2003

Texts in natural images contain significant and detailed informations about the images. Thus, to extract those texts correctly, we suggest a text extraction method using color variance feature. Generally, the texts in images have color variations with the backgrounds. Thus, if we express those variations in 3 dimensional RGB color space, we can emphasize the text regions that can be hard to be captured with a method using intensity variations in the gray-level images. We can even make robust extraction results with the images contaminated by light variations. The color variations are measured by color variance in this paper. First, horizontal and vertical variance images are obtained independently, and we can fine that the text regions have high values of the variances in both directions. Then, the two images are logically ANDed to remove the non-text components with only one directional high variance. We have applied the proposed method to the multiple kinds of the natural images, and we confirmed that the proposed feature can help to find the text regions that can he missed with the following features - intensity variations in the gray-level images and/or color continuity in the color images.

• Journal of KIISE:Software and Applications
• /
• v.31 no.4
• /
• pp.420-438
• /
• 2004

Artificially or naturally contained texts in the natural images have significant and detailed information about the scenes. If we develop a method that can extract and recognize those texts in real-time, the method can be applied to many important applications. In this paper, we suggest a new method that extracts the text areas in the natural images using the low-level image features of color continuity. gray-level variation and color valiance and that verifies the extracted candidate regions by using the high-level text feature such as stroke. And the two level features are combined hierarchically. The color continuity is used since most of the characters in the same text lesion have the same color, and the gray-level variation is used since the text strokes are distinctive in their gray-values to the background. Also, the color variance is used since the text strokes are distinctive in their gray-values to the background, and this value is more sensitive than the gray-level variations. The text level stroke features are extracted using a multi-resolution wavelet transforms on the local image areas and the feature vectors are input to a SVM(Support Vector Machine) classifier for the verification. We have tested the proposed method using various kinds of the natural images and have confirmed that the extraction rates are very high even in complex background images.

• Smart Media Journal
• /
• v.1 no.4
• /
• pp.27-34
• /
• 2012

In this paper, a novel text binarization is presented that can deal with some complex conditions, such as shadows, non-uniform illumination due to highlight or object projection, and messy backgrounds. To locate the target text region, a focus line is assumed to pass through a text region. Next, connected component analysis and stroke width estimation based on location information of the focus line is used to locate the bounding box of the text region, and each box of connected components. A series of classifications are applied to identify whether each CC(Connected component) is text or non-text. Also, a modified K-means clustering method based on an HCL color space is applied to reduce the color dimension. A text binarization procedure based on location of text component and seed color pixel is then used to generate the final result.

• The KIPS Transactions:PartB
• /
• v.13B no.3 s.106
• /
• pp.205-214
• /
• 2006

Texts in images have significant and detailed information about the scenes, and if we can automatically detect and recognize those texts in real-time, it can be used in various applications. In this paper, we propose a new text detection method that can find texts from the various camera-captured images and propose a text segmentation method from the detected text regions. The detection method proposes color variance as a detection feature in RGB color space, and the segmentation method suggests an improved K-means color clustering in RGB color space. We have tested the proposed methods using various kinds of document style and natural scene images captured by digital cameras and mobile-phone camera, and we also tested the method with a portion of ICDAR[1] contest images.

• Journal of Information Processing Systems
• /
• v.19 no.6
• /
• pp.842-857
• /
• 2023

Currently, most sentiment classification models on microblogging platforms analyze sentence parts of speech and emoticons without comprehending users' emotional inclinations and grasping moral nuances. This study proposes a hybrid sentiment analysis model. Given the distinct nature of microblog comments, the model employs a combined stop-word list and word2vec for word vectorization. To mitigate local information loss, the TextCNN model, devoid of pooling layers, is employed for local feature extraction, while BiLSTM is utilized for contextual feature extraction in deep learning. Subsequently, microblog comment sentiments are categorized using a classification layer. Given the binary classification task at the output layer and the numerous hidden layers within BiLSTM, the Tanh activation function is adopted in this model. Experimental findings demonstrate that the enhanced TextCNN-BiLSTM model attains a precision of 94.75%. This represents a 1.21%, 1.25%, and 1.25% enhancement in precision, recall, and F1 values, respectively, in comparison to the individual deep learning models TextCNN. Furthermore, it outperforms BiLSTM by 0.78%, 0.9%, and 0.9% in precision, recall, and F1 values.