• Journal of the Korean Institute of Intelligent Systems
• /
• v.21 no.2
• /
• pp.224-229
• /
• 2011

Collaborative filtering is used the most for recommendation systems because it can recommend potential items. However, when there are not many items to be evaluated, collaborative filtering can be subject to the influence of similarity or preference depending on the situation or the whim of the evaluator. In addition, by recommending items only on the basis of similarity with items that have been evaluated previously without relation to the present situation of the user, the recommendations become less accurate. In this paper, in order to solve the above problems, before starting the collaborative filtering procedure, we calculated similarity not by comparing all the values evaluated by users but rather by comparing only those users who were above the average in order to improve the accuracy of the recommendations. In addition, in the ceaselessly changing ubiquitous computing environment, it is not proper to recommend service information based only on the items evaluated by users. Therefore, we used methods of calculating similarity wherein the users' real time context information was used and a high weight was assigned to similar users. Such methods improved the recommendation accuracy by 16.2% on average.

• Smart Media Journal
• /
• v.8 no.4
• /
• pp.33-37
• /
• 2019

Because the ECG signals are time-series data acquired as time elapses, it is important to obtain comparative data the same in size as the enrolled data every time. This paper suggests a network model of GAN (Generative Adversarial Networks) based on an auxiliary classifier to generate synthetic ECG signals which may address the different data size issues. The Cosine similarity and Cross-correlation are used to examine the similarity of synthetic ECG signals. The analysis shows that the Average Cosine similarity was 0.991 and the Average Euclidean distance similarity based on cross-correlation was 0.25: such results indicate that data size difference issue can be resolved while the generated synthetic ECG signals, similar to real ECG signals, can create synthetic data even when the registered data are not the same as the comparative data in size.

• Journal of Korea Multimedia Society
• /
• v.14 no.11
• /
• pp.1467-1477
• /
• 2011

This paper proposes a HPPS(Hybrid Preference Prediction System) design using the analysis of user profile and of the similarity among users precisely to predict the preference for custom-tailored service. Contrary to the existing NBCFA(Neighborhood Based Collaborative Filtering Algorithm), this paper is designed using these following rules. First, if there is no neighbor's commodity rating value in a preference prediction formula, this formula uses the rating average value for a commodity. Second, this formula reflects the weighting value through the analysis of a user's characteristics. Finally, when the nearest neighbor is selected, we consider the similarity, the commodity rating, and the rating frequency. Therefore, the first and second preference prediction formula made HPPS improve the precision by 97.24%, and the nearest neighbor selection method made HPPS improve the precision by 75%, compared with the existing NBCFA.

• International Journal of Advanced Culture Technology
• /
• v.9 no.3
• /
• pp.265-270
• /
• 2021

VR and AR contents provide a rich user experience [1]. Realistic content with human computer interaction and immersion provides an improved user experience, but there is a limit to producing all elements realistically. In this study, we propose a method to advance the rendering of immersive content using background color information [2]. First, the elements necessary for Kelvin temperature rendering are derived from the color and background as context elements, and the rendering effect has been realized in the soap bubble. For soap bubbles Kelvin temperature rendering, the average color of the background is extracted and the color with the highest similarity is applied by comparing the main color and Kelvin temperature.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.16 no.2
• /
• pp.119-124
• /
• 2016

According to the expansion of smartphone penetration and development of wearable device, personal context information can be easily collected. To use this information, the context aware recommender system has been actively studied. The key issue in this field is how to deal with the context information, as users are influenced by different contexts while rating items. But measuring the similarity among contexts is not a trivial task. To solve this problem, we propose context aware post-filtering to apply the context compensation. To be specific, we calculate the compensation for different context information by measuring their average. After reflecting the compensation of the rating data, the mechanism recommends the items to the user. Based on the item recommendation list, we recover the rating score considering the context information. To verify the effectiveness of the proposed method, we use the real movie rating dataset. Experimental evaluation shows that our proposed method outperforms several state-of-the-art approaches.

• Journal of Information Science Theory and Practice
• /
• v.9 no.2
• /
• pp.1-17
• /
• 2021

Pseudo relevance feedback (PRF) is a powerful query expansion (QE) technique that prepares queries using the top k pseudorelevant documents and choosing expansion elements. Traditional PRF frameworks have robustly handled vocabulary mismatch corresponding to user queries and pertinent documents; nevertheless, expansion elements are chosen, disregarding similarity to the original query's elements. Word embedding (WE) schemes comprise techniques of significant interest concerning QE, that falls within the information retrieval domain. Deep averaging networks (DANs) defines a framework relying on average word presence passed through multiple linear layers. The complete query is understandably represented using the average vector comprising the query terms. The vector may be employed for determining expansion elements pertinent to the entire query. In this study, we suggest a DANs-based technique that augments PRF frameworks by integrating WE similarities to facilitate Arabic information retrieval. The technique is based on the fundamental that the top pseudo-relevant document set is assessed to determine candidate element distribution and select expansion terms appropriately, considering their similarity to the average vector representing the initial query elements. The Word2Vec model is selected for executing the experiments on a standard Arabic TREC 2001/2002 set. The majority of the evaluations indicate that the PRF implementation in the present study offers a significant performance improvement compared to that of the baseline PRF frameworks.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.2
• /
• pp.379-386
• /
• 2014

As the use of mobile device is increasing rapidly, the number of users is also increasing. However, most of the app stores are using recommendation of simple ranking method, so the accuracy of recommendation is lower. To recommend an item that is more appropriate to the user, this paper proposes a technique that reflects the weight of user information and recent preference degree of item. The proposed technique classifies the data set by categories and then derives a predicted value by applying the user's information weight to the collaborative filtering technique. To reflect the recent preference degree of item by categories, the average of items' rating values in the designated period is computed. An item is recommended by combining the two result values. The experiment result indicated that the proposed method has been more enhanced the accuracy, appropriacy, compared to item-based, user-based method.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.11
• /
• pp.2602-2608
• /
• 2013

User-based and item-based method have been developed as the solutions of the movie recommendation problem. However, these methods are faced with the sparsity problem and the problem of not reflecting user's rating respectively. In order to solve these problems, there is a research on the combination of the two methods using the concept of similarity. In reality, it is not free from the problem of sparsity, since it has a lot of parameters to be calculated. In this study, we propose a recommendation method using rating difference between items in order to complement this problem. This method is relatively free from the problem of sparsity, since it has less parameters to be calculated. And it can get more accurate results by reflecting the users rating to calculate the parameters. In experiments for the proposed method, the initial error is large, but the performance has been quickly stabilized after. In addition, it showed a 0.0538 lower average error compared to the existing method using similarity.

• Journal of Korea Multimedia Society
• /
• v.18 no.11
• /
• pp.1391-1399
• /
• 2015

In recent times, studies sentiment analysis are being actively conducted by implementing natural language processing technologies for analyzing subjective data such as opinions and attitudes of users expressed on the Web, blogs, and social networking services (SNSs). Conventionally, to classify the sentiments in texts, most studies determine positive/negative/neutral sentiments by assigning polarity values for sentiment vocabulary using sentiment lexicons. However, in this study, sentiments are classified based on Thayer's model, which is psychologically defined, unlike the polarity classification used in opinion mining. In this paper, as a method for classifying the sentiments, sentiment categories are proposed by extracting sentiment keywords for major sentiments by using hashtags, which are essential elements of Instagram. By applying sentiment categories to user posts, sentiments can be determined through the similarity measurement between the sentiment adjective candidates and the sentiment keywords. The test results of the proposed method show that the average accuracy rate for all the sentiment categories was 90.7%, which indicates good performance. If a sentiment classification system with a large capacity is prepared using the proposed method, then it is expected that sentiment analysis in various fields will be possible, such as for determining social phenomena through SNS.

• Proceedings of the IEEK Conference
• /
• 2002.06c
• /
• pp.227-230
• /
• 2002

In this paper we concentrate on overall direction of the gaze based on a head pose for human computer interaction. To decide a gaze direction of user in a image, it is important to pick up facial feature exactly. For this, we binarize the input image and search two eyes and the mouth through the similarity of each block ( aspect ratio, size, and average gray value ) and geometric information of face at the binarized image. We create a imaginary plane on the line made by features of the real face and the pin hole of the camera to decide the head orientation. We call it the virtual facial plane. The position of a virtual facial plane is estimated through projected facial feature on the image plane. We find a gaze direction using the surface normal vector of the virtual facial plane. This study using popular PC camera will contribute practical usage of gaze tracking technology.