• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.3
• /
• pp.1570-1594
• /
• 2017

Social network services have recently changed from environments for simply building connections among users to open platforms for generating and sharing various forms of information. Existing user reputation computation methods are inadequate for determining the trust in users on social media where explicit ratings are rare, because they determine the trust in users based on user profile, explicit relations, and explicit ratings. To solve this limitation of previous research, we propose a user reputation computation method suitable for the social media environment by incorporating implicit as well as explicit ratings. Reliable user reputation is estimated by identifying malicious information raters, modifying explicit ratings, and applying them to user reputation scores. The proposed method incorporates implicit ratings into user reputation estimation by differentiating positive and negative implicit ratings. Moreover, the method generates user reputation scores for individual categories to determine a given user's expertise, and incorporates the number of users who participated in rating to determine a given user's influence. This allows reputation scores to be generated also for users who have received no explicit ratings, and, thereby, is more suitable for social media. In addition, based on the user reputation scores, malicious information providers can be identified.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2005.05a
• /
• pp.629-638
• /
• 2005

In this paper, we consider two issues in collaborative filtering, which are closely related with the number of ratings of a user. First issue is the relationship between the number of ratings of a user and the performance of collaborative filtering. The relationship is investigated with two datasets, EachMovie and Movielens datasets. The number of ratings of a user is critical when the number of ratings is small, but after the number is over a certain threshold, its influence on recommendation performance becomes smaller. We also provide an explanation on the relationship between the number of ratings of a user and the performance in terms of neighborhood formations in collaborative filtering. The second issue is how to select an initial product list for new users for gaining user responses. We suggest and analyze 14 selection strategies which include popularity, favorite, clustering, genre, and entropy methods. Popularity methods are adequate for getting higher number of ratings from users, and favorite methods are good for higher average preference ratings of users.

• ETRI Journal
• /
• v.43 no.1
• /
• pp.95-108
• /
• 2021

Application (app) ratings are feedback provided voluntarily by users and serve as important evaluation criteria for apps. However, these ratings can often be biased owing to insufficient or missing votes. Additionally, significant differences have been observed between numeric ratings and user reviews. This study aims to predict the numeric ratings of Google apps using machine learning classifiers. It exploits numeric app ratings provided by users as training data and returns authentic mobile app ratings by analyzing user reviews. An ensemble learning model is proposed for this purpose that considers term frequency/inverse document frequency (TF/IDF) features. Three TF/IDF features, including unigrams, bigrams, and trigrams, were used. The dataset was scraped from the Google Play store, extracting data from 14 different app categories. Biased and unbiased user ratings were discriminated using TextBlob analysis to formulate the ground truth, from which the classifier prediction accuracy was then evaluated. The results demonstrate the high potential for machine learning-based classifiers to predict authentic numeric ratings based on actual user reviews.

• Journal of Information Processing Systems
• /
• v.17 no.3
• /
• pp.462-472
• /
• 2021

With an increase in the scale of recommender systems, users' rating data tend to be extremely sparse. Some methods have been utilized to alleviate this problem; nevertheless, it has not been satisfactorily solved yet. Therefore, we propose an effective pre-rating method based on users' dichotomous preferences and average ratings fusion. First, based on a user-item ratings matrix, a new user-item preference matrix was constructed to analyze and model user preferences. The items were then divided into two categories based on a parameterized dynamic threshold. The missing ratings for items that the user was not interested in were directly filled with the lowest user rating; otherwise, fusion ratings were utilized to fill the missing ratings. Further, an optimized parameter λ was introduced to adjust their weights. Finally, we verified our method on a standard dataset. The experimental results show that our method can effectively reduce the prediction error and improve the recommendation quality. As for its application, our method is effective, but not complicated.

• Journal of Information Technology Services
• /
• v.16 no.4
• /
• pp.161-176
• /
• 2017

Recently, most of the users can easily get access to a variety of information sources about companies, products, and services through online channels. Therefore, the online user evaluations are becoming the most powerful tool to generate word of mouth. The user's evaluation is provided in two forms, quantitative rating and review text. The rating is then divided into an overall rating and a detailed rating according to various evaluation criteria. However, since it is a burden for the reviewer to complete all required ratings for each evaluation criteria, so most of the sites requested only mandatory inputs for overall rating and optional inputs for other evaluation criteria. In fact, many users input only the ratings for some of the evaluation criteria and the percentage of missed ratings for each criteria is about 40%. As these missed ratings are the missing values in each criteria, the simple average calculation by ignoring the average 40% of the missed ratings can sufficiently distort the actual phenomenon. Therefore, in this study, we propose a methodology to predict the rating for the missed values of each criteria by analyzing user's evaluation information included the overall rating and text review for each criteria. The experiments were conducted on 207,968 evaluations collected from the actual hotel evaluation site. As a result, it was confirmed that the prediction accuracy of the detailed criteria ratings by the proposed methodology was much higher than the existing average-based method.

• Proceedings of the Korean Information Science Society Conference
• /
• 2003.04c
• /
• pp.386-388
• /
• 2003

We are developing a web-based movie recommender system that catches and reasons with user profiles and ratings to recommend movies. In the paper, we outline the current status of our implementation with particular emphasis on the mechanisms used to provide effective recommendations. Social recommender systems collect ratings of items from many individuals and use nearest-neighbor techniques to make recommendations to a user. However, these methods only depend on the ratings and ignore other useful information. Our primary concern is to provide an approach that can recommend the movies based on not only the user ratings but also the significant amount of other information that is available about the nature of each items - such as cast list or movie genre. We experimentally evaluate our approach and compare them to conventional social filtering, which suggests merits to our approach.

• Journal of Information Processing Systems
• /
• v.17 no.4
• /
• pp.707-720
• /
• 2021

To improve the accuracy of the recommendation system, multi-criteria recommendation systems have been widely researched. However, it is highly complicated to extract the preferred features of users and items from the data. To this end, subjective indicators, which indicate a user's priorities for personalized recommendations, should be derived. In this study, we propose a method for generating recommendation candidates by predicting multi-criteria ratings from reviews and using them to derive user priorities. Using a deep learning model based on convolutional neural network (CNN) and bidirectional long short-term memory (BiLSTM), multi-criteria prediction ratings were derived from reviews. These ratings were then aggregated to form a linear regression model to predict the overall rating. This model not only predicts the overall rating but also uses the training weights from the layers of the model as the user's priority. Based on this, a new score matrix for recommendation is derived by calculating the similarity between the user and the item according to the criteria, and an item suitable for the user is proposed. The experiment was conducted by collecting the actual "TripAdvisor" dataset. For performance evaluation, the proposed method was compared with a general recommendation system based on singular value decomposition. The results of the experiments demonstrate the high performance of the proposed method.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.11
• /
• pp.5321-5334
• /
• 2019

Data mining technology is frequently used in identifying the intention of users over a variety of information contexts. Since relevant terms are mainly hidden in text data, it is necessary to extract them. Quantification is required in order to interpret user preference in association with other structured data. This paper proposes rating and comments mining to identify user priority and obtain improved ratings. Structured data (location and rating) and unstructured data (comments) are collected and priority is derived by analyzing statistics and employing TF-IDF. In addition, the improved ratings are generated by applying priority categories based on materialized ratings through Sentiment-Oriented Point-wise Mutual Information (SO-PMI)-based emotion analysis. In this paper, an experiment was carried out by collecting ratings and comments on "place" and by applying them. We confirmed that the proposed mining method is 1.2 times better than the conventional methods that do not reflect priorities and that the performance is improved to almost 2 times when the number to be predicted is small.

• The KIPS Transactions:PartB
• /
• v.14B no.7
• /
• pp.549-558
• /
• 2007

Collaborative filtering systems have problems when users rate items and the rated results depend on their feelings, as there is a possibility that the results include noise. The method proposed in this paper optimizes the matrix by excluding irrelevant ratings as information for recommendations from a user-item matrix using dispersion. It reduces the noise that results from predicting preferences based on original user ratings by inflecting the information for items and users on the matrix. The method excludes the ratings values of the utmost limits using a percentile to supply the defects of coefficient of variance and composes a weighted user-item matrix by combining the user coefficient of variance with the median of ratings for items. Finally, the preferences of the active user are predicted based on the weighted matrix. A large database of user ratings for movies from the MovieLens recommender system is used, and the performance is evaluated. The proposed method is shown to outperform earlier methods significantly.

• The Journal of Korean Association of Computer Education
• /
• v.14 no.5
• /
• pp.97-104
• /
• 2011

Collaborative filtering is a methodology to recommend websites by obtaining data and opinions from the other users with similar tastes. During the past few years, this method has been used in various fields such as books, food, and movies in e-commerce systems. This study addresses the computation of similarity between users to determine items to be recommended in collaborative filtering systems. Previous studies measured similarity between users by treating each user's ratings independently without considering the distribution of the user's ratings. In contrast, this study measures similarity by utilizing position and rank information of each rating in the range of the user's ratings. The result of the experiments on the real datasets demonstrated that the proposed method improves the mean absolute error significantly, compared to the previous methods, especially when the predetermined range of ratings is large.