• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.11
• /
• pp.535-540
• /
• 2021

Multi-document refers to a document consisting of various topics, not a single topic, and a typical example is online reviews. There have been several attempts to summarize online reviews because of their vast amounts of information. However, collective summarization of reviews through existing summary models creates a problem of losing the various topics that make up the reviews. Therefore, in this paper, we present method to summarize the review with minimal loss of the topic. The proposed method classify reviews through processes such as preprocessing, importance evaluation, embedding substitution using BERT, and embedding clustering. Furthermore, the classified sentences generate the final summary using the trained Transformer summary model. The performance evaluation of the proposed model was compared by evaluating the existing summary model, seq2seq model, and the cosine similarity with the ROUGE score, and performed a high performance summary compared to the existing summary model.

• The Journal of the Korea Contents Association
• /
• v.22 no.5
• /
• pp.1-15
• /
• 2022

In this study, text analysis was performed on the mealkit product review data to identify factors affecting the evaluation of the mealkit product. The data used for the analysis were collected by scraping 334,498 reviews of mealkit products in Naver shopping site. After preprocessing the text data, wordclouds and sentiment analyses based on word frequency and normalized TF-IDF were performed. Logistic regression model was applied to predict the polarity of reviews on mealkit products. From the logistic regression models derived for each product category, the main factors that caused positive and negative emotions were identified. As a result, it was verified that text analysis can be a useful tool that provides a basis for maximizing positive factors for a specific category, menu, and material and removing negative risk factors when developing a mealkit product.

• Journal of Internet Computing and Services
• /
• v.21 no.5
• /
• pp.1-7
• /
• 2020

Deep neural networks(DNN), which are used as approximation functions in reinforcement learning (RN), theoretically can be attributed to realistic results. In empirical benchmark works, time difference learning (TD) shows better results than Monte-Carlo learning (MC). However, among some previous works show that MC is better than TD when the reward is very rare or delayed. Also, another recent research shows when the information observed by the agent from the environment is partial on complex control works, it indicates that the MC prediction is superior to the TD-based methods. Most of these environments can be regarded as 5-step Q-learning or 20-step Q-learning, where the experiment continues without long roll-outs for alleviating reduce performance degradation. In other words, for networks with a noise, a representative network that is regardless of the controlled roll-outs, it is better to learn MC, which is robust to noisy rewards than TD, or almost identical to MC. These studies provide a break with that TD is better than MC. These recent research results show that the way combining MC and TD is better than the theoretical one. Therefore, in this study, based on the results shown in previous studies, we attempt to exploit a random balance with a mixture of TD and MC in RL without any complicated formulas by rewards used in those studies do. Compared to the DQN using the MC and TD random mixture and the well-known DQN using only the TD-based learning, we demonstrate that a well-performed TD learning are also granted special favor of the mixture of TD and MC through an experiments in OpenAI Gym.

• Journal of the Korea Convergence Society
• /
• v.11 no.10
• /
• pp.187-195
• /
• 2020

The purpose of study is to find ways to improve online research work in the near future environment. Study methods and results are as follows. First, the basis of research was secured through theoretical consideration. Second, three types were derived by collecting examples. There are reward, execution and visualization. Third, comprehensive matrix and directions were derived through case analysis. According to the study, the most frequently used phase of AI technology were analysis, visualization, execution, casting, design, and matching, The direction of use of AI can be largely divided into work efficiency, human thought assistance, and changes in research methods. But the plan phase is generally scarce. Overall, the lack of in function is provided, including AI technology. At a time when research plans are becoming more important as users and products diversify.

• The Journal of the Korea Contents Association
• /
• v.20 no.12
• /
• pp.13-22
• /
• 2020

This study aims to find out success factors of crowdfunding on movie projects. For empirical analysis, we collected 583 data of the movie projects from the crowdfunding platform 'Tumblbug'. To figure out the success factors, we examined effects of 10 independent variables on 1 dependent variable. The independent variable includes target amount, project information, reward options, creator funding power, editor recommendation, creator contents power, movie type, number of comments, number of replies, and number of SNS information. The final achievement rate of crowdfunding was set as dependent variable. This study found that the target amount, number of text information, number of video information, editor recommendation, number of backers' reply, and number of SNS information had a significant impact on the achievement rate of the movie crowdfunding project. This study has implications in that it has discovered a variable of editor recommendation and the number of SNS information, and both of them have a positive effect on crowdfunding achievement.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.1
• /
• pp.31-40
• /
• 2022

This paper proposes an augmented QSBC(Quantum Short-Block Code) that preserves the function of the existing QSBC and adds a single bit-flip error correction function due to Pauli X and Y errors. The augmented QSBC provides the diagnosis and automatic correction of a single Pauli X error by inserting additional auxiliary qubits and Toffoli gates as many as the number of information words into the existing QSBC. In this paper, the general expansion method of the augmented QSBC using seed vector and the realization method of the Toffoli gate of the single bit-flip error automatic correction function reflecting the scalability are also presented. The augmented QSBC proposed in this paper has a trade-off with a coding rate of at least 1/3 and at most 1/2 due to the insertion of auxiliary qubits.

• Journal of the Korea Computer Graphics Society
• /
• v.29 no.4
• /
• pp.7-16
• /
• 2023

For decades, creating a desired locomotive motion in a goal-oriented manner has been a challenge in character animation. Data-driven methods using generative models have demonstrated efficient ways of predicting long sequences of motions without the need for explicit conditioning. While these methods produce high-quality long-term motions, they can be limited when it comes to synthesizing motion for challenging novel scenarios, such as punching a random target. A state-of-the-art solution to overcome this limitation is by using a GAN Discriminator to imitate motion data clips and incorporating reinforcement learning to compose goal-oriented motions. In this paper, our research aims to create characters performing combat sports such as boxing, using a novel reward design in conjunction with existing GAN-based approaches. We experimentally demonstrate that both the Adversarial Motion Prior [3] and Adversarial Skill Embeddings [4] methods are capable of generating viable motions for a character punching a random target, even in the absence of mocap data that specifically captures the transition between punching and locomotion. Also, with a single learned policy, multiple task controllers can be constructed through the TimeChamber framework.

• Journal of Korea Multimedia Society
• /
• v.19 no.5
• /
• pp.948-956
• /
• 2016

Many franchise stores are in the surrounding. These stores offer many membership services. But, small traders do not provide much customer management service. For example, non-franchised cafes provide the services through paper coupons. However, most of paper coupons are available only in one store that issued the coupon. Besides, these coupons are in stamp format. Due to the absence of customers' management service, it is hard for small traders to attract customers compared to the franchise. Therefore, in this paper, We had implemented a mobile application, and applied to new customer management service in order to increase the price competitiveness of small traders. This service issues electronic coupon through a mobile app. customers can receive a relatively large amount of discount, and a deadline of coupons are short. When customers use coupons, new coupons are issued at the same time. This structure can be powerful means to lead customers' frequent visit. Small traders can gain a lot of regular customers by using this service.

• The Journal of Korean Association of Computer Education
• /
• v.17 no.6
• /
• pp.115-122
• /
• 2014

One of the problems with current youth generations is increasing rate of violence and suicide in their school lives, and this study aims at the design of a sentiment analysis system to prevent suicide by uising big data process. The main issues of the design are economical implementation, easy and fast processing for the users, so, the open source Hadoop system with MapReduce algorithm is used on the HDFS(Hadoop Distributed File System) for the experimentation. This study uses word count method to do the sentiment analysis with informal data on some sns communications concerning a kinds of violent words, in terms of text mining to avoid some expensive and complex statistical analysis methods.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.9 no.6
• /
• pp.605-614
• /
• 1999

This paper addresses analysis and design of a class of complex single-input single-output fuzzy control systems. In the proposed method, the fuzzy model, which represents the local dynamic behavior of the given nonlinear system, is utilized to construct the controller. The overall controller consists of the local compensators which compensate the local dynamic linear model and the feed-forward controller which is designed via sliding mode control theory. Therefore, the globally stable fuzzy controller is designed without finding a common Lyapunov matrix. and shows improved perfonnance and tracking results by taking the advantages of fuzzy-model-based control theory and sliding mode control theory. Furthennore, stability analysis is conducted not Ibr the fuzzy model but for the real underlying nonlinear system. Two numerical examples are included to show the effcctiveness and feasibility of the proposed fuzzy control method.