• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.319-320
• /
• 2023

일반적인 여행 커뮤니티는 사진과 텍스트 기반의 사용자 리뷰를 바탕으로 정보 공유를 한다. 본 연구에서는 관광지에서 수행한 활동을 한 문장의 형태로 공유하는 애플리케이션을 제안한다. ChatGPT를 활용하여 활동을 산책, 사진, 음식 등 9가지 태그로 분류하여 관광지가 가지는 특징을 용이하게 파악한다. 또한, 사용자가 작성한 활동을 임베딩하고 관광지 소개 글 벡터와 유사도를 비교하여 관광지를 추천한다. 본 애플리케이션을 통해 사용자가 긴 설명이나 사진 없이 관광지가 가지는 정보를 쉽게 공유하고 관광지 추천을 하는 새로운 여행 커뮤니티를 제공할 수 있을 것으로 기대한다.

• The KIPS Transactions:PartB
• /
• v.17B no.1
• /
• pp.85-92
• /
• 2010

A shallow semantic parsing system analyzes the relationship that a syntactic constituent of the sentence has with a predicate. It identifies semantic arguments representing agent, patient, instrument, etc. of the predicate. In this study, we propose a two-phase shallow semantic parsing model which consists of the identification phase and the classification phase. We first find the boundary of semantic arguments from partial syntactic parsing results, and then assign appropriate semantic roles to the identified semantic arguments. By taking the sequential two-phase approach, we can alleviate the unbalanced class distribution problem, and select the features appropriate for each task. Experiments show the relative contribution of each phase on the test data.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.12 no.5
• /
• pp.2402-2409
• /
• 2011

The voice has distinguishable acoustic properties of speaker as well as transmitting information. The speaker recognition is the method to figures out who speaks the words through acoustic differences between speakers. The speaker recognition is roughly divided two kinds of categories: speaker verification and identification. The speaker verification is the method which verifies speaker himself based on only one's voice. Otherwise, the speaker identification is the method to find speaker by searching most similar model in the database previously consisted of multiple subordinate sentences. This paper composes feature vector from extracting MFCC coefficients and uses the dynamic time warping algorithm to compare the similarity between features. In order to describe common characteristic based on phonological features of spoken words, two subordinate sentences for each speaker are used as the training data. Thus, it is possible to identify the speaker who didn't say the same word which is previously stored in the database.

• Korean Journal of Cognitive Science
• /
• v.27 no.1
• /
• pp.43-60
• /
• 2016

Recently, Unsupervised Word Sense Disambiguation research has focused on Graph based disambiguation. Graph-based disambiguation has built a semantic graph based on words collocated in context or sentence. However, building such a graph over all ambiguous word lead to unnecessary addition of edges and nodes (and hence increasing the error). In contrast, our work uses Word2Vec to consider the most similar words to an ambiguous word in the context or sentences, to rebuild a graph of the matched words. As a result, we show a higher F1-Measure value than the previous methods by using Word2Vec.

• Korean Journal of Cognitive Science
• /
• v.12 no.1_2
• /
• pp.35-42
• /
• 2001

In this paper. a document is represented as a weighted graph called a text relationship map. In the graph. a node represents a vector of nouns in a sentence, an edge completely connects other nodes. and a weight on the edge is a value of the similarity between two nodes. The similarity is based on the word overlap between the corresponding nodes. The importance of a node. called an aggregate similarity in this paper. is defined as the sum of weights on the links connecting it to other nodes on the map. In this paper. we present a Korean text summarization system using the aggregate similarity. To evaluate our system, we used two test collection, one collection (PAPER-InCon) consists of 100 papers in the field of computer science: the other collection (NEWS) is composed of 105 articles in the newspapers and had built by KOROlC. Under the compression rate of 20%. we achieved the recall of 46.6% (PAPER-InCon) and 30.5% (NEWS) and the precision of 76.9% (PAPER-InCon) and 42.3% (NEWS).

• Journal of Intelligence and Information Systems
• /
• v.28 no.2
• /
• pp.127-146
• /
• 2022

Recently, as word embedding has shown excellent performance in various tasks of deep learning-based natural language processing, researches on the advancement and application of word, sentence, and document embedding are being actively conducted. Among them, cross-language transfer, which enables semantic exchange between different languages, is growing simultaneously with the development of embedding models. Academia's interests in vector alignment are growing with the expectation that it can be applied to various embedding-based analysis. In particular, vector alignment is expected to be applied to mapping between specialized domains and generalized domains. In other words, it is expected that it will be possible to map the vocabulary of specialized fields such as R&D, medicine, and law into the space of the pre-trained language model learned with huge volume of general-purpose documents, or provide a clue for mapping vocabulary between mutually different specialized fields. However, since linear-based vector alignment which has been mainly studied in academia basically assumes statistical linearity, it tends to simplify the vector space. This essentially assumes that different types of vector spaces are geometrically similar, which yields a limitation that it causes inevitable distortion in the alignment process. To overcome this limitation, we propose a deep learning-based vector alignment methodology that effectively learns the nonlinearity of data. The proposed methodology consists of sequential learning of a skip-connected autoencoder and a regression model to align the specialized word embedding expressed in each space to the general embedding space. Finally, through the inference of the two trained models, the specialized vocabulary can be aligned in the general space. To verify the performance of the proposed methodology, an experiment was performed on a total of 77,578 documents in the field of 'health care' among national R&D tasks performed from 2011 to 2020. As a result, it was confirmed that the proposed methodology showed superior performance in terms of cosine similarity compared to the existing linear vector alignment.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.12
• /
• pp.988-992
• /
• 2009

In this paper, we attempted to extract binary relations between terminologies using composite kernels consisting of convolution parse tree kernels and WordNet verb synset vector kernels which explain the semantic relationships between two entities in a sentence. In order to evaluate the performance of our system, we used three domain specific test collections. The experimental results demonstrate the superiority of our system in all the targeted collection. Especially, the increase in the effectiveness on KREC 2008, 8% in F1, shows that the core contexts around the entities play an important role in boosting the entire performance of relation extraction.

• Journal of KIISE
• /
• v.44 no.12
• /
• pp.1290-1295
• /
• 2017

Document classification, a task of classifying the category of each document based on text, is one of the fundamental areas for natural language processing. Document classification may be used in various fields such as topic classification and sentiment classification. Neural network models for document classification can be divided into two categories: word-level models and character-level models that treat words and characters as basic units respectively. In this study, we propose a neural network model that combines character-level and word-level models to improve performance of document classification. The proposed model extracts the feature vector of each word by combining information obtained from a word embedding matrix and information encoded by a character-level neural network. Based on feature vectors of words, the model classifies documents with a hierarchical structure wherein recurrent neural networks with attention mechanisms are used for both the word and the sentence levels. Experiments on real life datasets demonstrate effectiveness of our proposed model.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.11
• /
• pp.41-50
• /
• 2020

Sentimental analysis begins with the search for words that determine the sentimentality inherent in data. Managers can understand market sentimentality by analyzing a number of relevant sentiment words which consumers usually tend to use. In this study, we propose exploring performance of feature selection methods embedded with Particle Swarm Optimization Multi Objectives Evolutionary Algorithms. The performance of the feature selection methods was benchmarked with machine learning classifiers such as Decision Tree, Naive Bayesian Network, Support Vector Machine, Random Forest, Bagging, Random Subspace, and Rotation Forest. Our empirical results of opinion mining revealed that the number of features was significantly reduced and the performance was not hurt. In specific, the Support Vector Machine showed the highest accuracy. Random subspace produced the best AUC results.

• Journal of KIISE
• /
• v.44 no.7
• /
• pp.742-747
• /
• 2017

In this paper, we propose a novel approach to improve the performance of the Convolutional Neural Network(CNN) word embedding model on top of word2vec with the result of performing like doc2vec in conducting a document classification task. The Word Piece Model(WPM) is empirically proven to outperform other tokenization methods such as the phrase unit, a part-of-speech tagger with substantial experimental evidence (classification rate: 79.5%). Further, we conducted an experiment to classify ten categories of news articles written in Korean by feeding words and document vectors generated by an application of WPM to the baseline and the proposed model. From the results of the experiment, we report the model we proposed showed a higher classification rate (89.88%) than its counterpart model (86.89%), achieving a 22.80% improvement. Throughout this research, it is demonstrated that applying doc2vec in the document classification task yields more effective results because doc2vec generates similar document vector representation for documents belonging to the same category.