• The KIPS Transactions:PartB
• /
• v.11B no.4
• /
• pp.509-516
• /
• 2004

Semantic interpretation of image is incomplete without some mechanism for understanding semantic content that is not directly visible. For this reason, human assisted content-annotation through natural language is an attachment of textual description to image. However, keyword-based retrieval is in the level of syntactic pattern matching. In other words, dissimilarity computation among terms is usually done by using string matching not concept matching. In this paper, we propose a method for computerized semantic similarity calculation In WordNet space. We consider the edge, depth, link type and density as well as existence of common ancestors. Also, we have introduced method that applied similarity measurement on semantic image retrieval. To combine wi#h the low level features, we use the spatial color distribution model. When tested on a image set of Microsoft's 'Design Gallery Line', proposed method outperforms other approach.

• Phonetics and Speech Sciences
• /
• v.4 no.4
• /
• pp.45-52
• /
• 2012

The purpose of this paper is to measure the first two vowel formants of the ten young female speakers from the Buckeye Corpus of Conversational Speech [1] automatically and then to analyze various potential factors that may affect the formant distribution of the eight peripheral vowels of English. The factors that were analyzed included the place of articulation, the content versus function word information, the syllabic stress information, the location in a word, the location in an utterance, the speech rate of the three consecutive words, and the word frequency in the corpus. The results indicate that the overall formant patterns of the female speakers were similar to those of earlier works. The effects of the factors on the realization of the two formants were also similar to those from the male speakers with minor differences.

• The Journal of Industrial Distribution & Business
• /
• v.11 no.8
• /
• pp.7-19
• /
• 2020

Purpose: Prior studies rarely investigated the effects of the Word of Mouse (WoM) information on consumers' motivation and willingness to purchase a product. Furthermore, few scholars have studied how word-of-mouth information works and they fail to find consistent results. Research design, data and methodology: This study uses a multivariate regression model to investigate the influence of WoM on consumer attitudes and consumer decision-making. It categorizes the quality of WoM into source level and acceptance level, to analyze its influence from a new perspective. A total of 400 surveys were completed, resulting in 336 usable questionnaires for analysis. It was collected in 14 cities from all regions in China. This study constructs a theoretical model of WoM influence on consumers' purchase willingness based on a systematic review of the related literature on WoM quality, perceived value, customer trust, and consumers' purchase willingness. Results: Empirical results reveal that the Internet WoM (consumer's source level and acceptance level) indirectly affects consumer behavior by influencing consumer attitudes. Conclusions: This study provides practical significance and value for merchants to develop better WoM marketing and to establish the reliability of WoM websites. Companies should consider online WoM from the perspective of consumers, thereby improving existing marketing strategies.

• Journal of Distribution Science
• /
• v.18 no.2
• /
• pp.39-48
• /
• 2020

Purpose: This study examines components that dictate service quality of Online Travel Agencies. Research design, data and methodology: This study performed a survey that targeted people who have purchased travel products using Online Travel Agencies in the past year at Incheon International Airport. Out of 280 questionnaires, this study selected 249(88.9%) questionnaires for analysis. In analysis, this study used statistical package called 'SPSS 22'. Results: Based on the results, this study identified that company reputation and transactional stability influenced consumers' intent to share anecdotal and practical information. Conclusions: This study discovered several major findings. First, 'convenience' and 'price', two major factors in service quality, imposed positive effects on word-of-mouth intention of consumers. In particular, 'convenience' had the most significant impact. Next, this study verified the moderating effects of 'transaction safety' and 'reputation' on the effect relationships between 'convenience' and 'price', and consumer word-of-mouth intention. The results conclude that, in fact, some statistically significant differences among the effect degrees of the two independent variables on the dependent variable according to the safety level of transaction on Online Travel Agencies website do exist. In short, 'transaction safety' variable does have moderating effects on the above relationships.

• Journal of Information Processing Systems
• /
• v.16 no.2
• /
• pp.329-342
• /
• 2020

In recent years, emotional text classification is one of the essential research contents in the field of natural language processing. It has been widely used in the sentiment analysis of commodities like hotels, and other commentary corpus. This paper proposes an improved W-LDA (weighted latent Dirichlet allocation) topic model to improve the shortcomings of traditional LDA topic models. In the process of the topic of word sampling and its word distribution expectation calculation of the Gibbs of the W-LDA topic model. An average weighted value is adopted to avoid topic-related words from being submerged by high-frequency words, to improve the distinction of the topic. It further integrates the highest classification of the algorithm of support vector machine based on the extracted high-quality document-topic distribution and topic-word vectors. Finally, an efficient integration method is constructed for the analysis and extraction of emotional words, topic distribution calculations, and sentiment classification. Through tests on real teaching evaluation data and test set of public comment set, the results show that the method proposed in the paper has distinct advantages compared with other two typical algorithms in terms of subject differentiation, classification precision, and F1-measure.

• Journal of the Korean Society of Clothing and Textiles
• /
• v.35 no.7
• /
• pp.787-800
• /
• 2011

This study investigated service recovery justice perception, purchase satisfaction, positive word-of- mouth (WOM) intention, and repurchase intention of fashion consumers according to the types of internet shopping malls. A survey was conducted from December 20 in 2010 to January 28 in 2011, and 324 respondents who had complaint with internet shopping malls were used in the data analysis. The statistical analysis methods were frequency analysis, factor analysis, reliability analysis, t-test, ANOVA, and multiple regression analysis. The results of this study were as follows. First, in case of integrated internet shopping malls, the procedure justice influenced the purchase satisfaction and the interaction justice influenced the positive WOM intention. In addition, the distribution justice influenced the repurchase intention of fashion consumers. In the case of an open market, the interaction justice influenced the purchase satisfaction and the distribution justice influenced the positive WOM intention. In case of specialized internet shopping malls, the distribution justice influenced the purchase satisfaction and the interaction justice influenced the repurchase intention. Second, there was a difference in the procedure justice, positive WOM intention, and repurchase intention according to gender.

• The Journal of Asian Finance, Economics and Business
• /
• v.8 no.3
• /
• pp.793-805
• /
• 2021

Corporate Social Responsibility (CSR) program has become one of the primary concerns of companies worldwide. For many companies, treating the environment and the community well is important to business practice and reputation, and this is reflected in their CSR programs. CSR is a company's obligation to consider the interests of its employees, customers, shareholders, communities, and the environment and to consider the social and environmental consequences of their business activities. CSR plays an important role in relationship building with customers. Thus, the main purpose of this study is to analyze the relationship between customer perception of CSR and customer trust. The model of this study considers two mediating variables, i.e., company reputation and word of mouth to link CSR to customer trust. This study employs a causal survey design. The respondents were 160 consumers who have purchased products and knew about CSR programs of a global food company. Data analysis was using structural equation modeling (SEM) to test the hypotheses. The results of this study revealed that CSR negatively impacts customer trust. Furthermore, CSR positively impacts corporate reputation and word of mouth. Besides, this study found corporate reputation positively impacts word of mouth and customer trust. While the mediating effect of reputation and word of mouth also positively impacts the relationship between CSR and consumer trust. A good reputation and word of mouth could be connecting buyers and enhances the power of suppliers.

• Phonetics and Speech Sciences
• /
• v.6 no.3
• /
• pp.155-164
• /
• 2014

This paper proposes a new method to determine the recognition units for large vocabulary continuous speech recognition (LVCSR) in Korean by applying unsupervised segmentation and merging. In the proposed method, a text sentence is segmented into morphemes and position information is added to morphemes. Then submorpheme units are obtained by splitting the morpheme units through the maximization of posterior probability terms. The posterior probability terms are computed from the morpheme frequency distribution, the morpheme length distribution, and the morpheme frequency-of-frequency distribution. Finally, the recognition units are obtained by sequentially merging the submorpheme pair with the highest frequency. Computer experiments are conducted using a Korean LVCSR with a 100k word vocabulary and a trigram language model obtained by a 300 million eojeol (word phrase) corpus. The proposed method is shown to reduce the out-of-vocabulary rate to 1.8% and reduce the syllable error rate relatively by 14.0%.

• The Journal of Asian Finance, Economics and Business
• /
• v.7 no.11
• /
• pp.449-457
• /
• 2020

This study's purpose was to empirically research the effect of brand image and brand love on brand commitment, and positive word of mouth. We analyzed research data from 267 customers at electronics stores in Ho Chi Minh City, Vietnam. The convenient sampling method was used to collect the research sample. The measurement applied a 5-point Likert scale classifying from 1= completely disagree and 5=completely agree. Based on previous studies, the research model was recommended. The Partial Least Squares method was done to examine the measurement model and the structural model. We had estimated the reliability of the scales through Cronbach's alpha and composite reliability. Besides, we had assessed discriminant validity through the Fornell-Larcker standard. The outcome of the study illustrated that brand image had a significantly positive influence on brand love. Besides, the results of the research declared that brand image was positively related to brand commitment. The results also revealed that brand image was positively linked to positive word of mouth. Likewise, the findings of the study disclosed that brand love had a positive impact on brand commitment and positive word of mouth. Additionally, the results of the research confirmed that brand commitment was positively linked to word of mouth.

• Journal of Korea Multimedia Society
• /
• v.24 no.10
• /
• pp.1391-1402
• /
• 2021

This paper focuses use context-sensitive spelling error correction using generative adversarial network. Generative adversarial network[1] are attracting attention as they solve data generation problems that have been a challenge in the field of deep learning. In this paper, sentences are generated using word embedding information and reflected in word distribution representation. We experiment with DCGAN[2] used for the stability of learning in the existing image processing and D2GAN[3] with double discriminator. In this paper, we experimented with how the composition of generative adversarial networks and the change of learning corpus influence the context-sensitive spelling error correction In the experiment, we correction the generated word embedding information and compare the performance with the actual word embedding information.