Content-based Recommendation Methods Research Articles

Paper Recommendation Method based on Attention Mechanism and Graph Neural Network

At present, most recommendation technologies only consider text or citation information, which suffers from data sparseness and cold start problems. Therefore, an academic paper recommendation method based on attention mechanism and heterogeneous graph CAH is proposed. This method considers textual information and heterogeneous graph structure information to obtain a richer and more complete feature representation. Finally, cosine similarity is calculated to generate recommendations. The results show that compared with the content-based recommendation method, the accuracy rate, recall rate and f value of CAH method are increased by nearly 5.6%, 5.8% and 8.7%, respectively, which are significantly improved compared with the basic method. This method is expected to promote the in-depth application of recommendation systems in the field of artificial intelligence.

Knowledge-Aware Explainable Reciprocal Recommendation

Reciprocal recommender systems (RRS) have been widely used in online platforms such as online dating and recruitment. They can simultaneously fulfill the needs of both parties involved in the recommendation process. Due to the inherent nature of the task, interaction data is relatively sparse compared to other recommendation tasks. Existing works mainly address this issue through content-based recommendation methods. However, these methods often implicitly model textual information from a unified perspective, making it challenging to capture the distinct intentions held by each party, which further leads to limited performance and the lack of interpretability. In this paper, we propose a Knowledge-Aware Explainable Reciprocal Recommender System (KAERR), which models metapaths between two parties independently, considering their respective perspectives and requirements. Various metapaths are fused using an attention-based mechanism, where the attention weights unveil dual-perspective preferences and provide recommendation explanations for both parties. Extensive experiments on two real-world datasets from diverse scenarios demonstrate that the proposed model outperforms state-of-the-art baselines, while also delivering compelling reasons for recommendations to both parties.

Session-Based Recommendation with GNN and Time-Aware Memory Network

The goal of the session-based recommendation system (SBRS) is to predict the user’s next behavior based on anonymous sessions. Since long-term historical information of users is not available, deep learning technology has become the mainstream technology in session-based recommendation systems instead of traditional content-based recommendation methods. However, most SBRS methods only consider the session itself, ignoring the collaborative information from other sessions. Even if some SBRS models consider collaborations between sessions, they mostly use the click order to calculate the similarity only and ignore the time the user spends on different items, which might imply the user’s varying interest on these items. In this paper, we propose a session-based recommendation model with GNN and time-aware memory networks (SR-GTM), which learns the user’s interest representation by combining the information from the session itself and the collaborative information from relevant neighbor sessions. Specifically, SR-GTM mainly includes inner feature extraction module (IFEM) and outer feature extraction module (OFEM). IFEM uses GNN to learn the session features based on its item sequence, and OFEM uses a memory network with dwell time information encoded to extract collaborative information. Finally, SR-GTM aggregates IFEM and OFEM by the gating mechanism and then decodes the output by a softmax layer to obtain the recommendation score for each candidate item. Experiments on three public datasets Yoochoose1/64, Yoochoose1/4, and RetailRocket show that SR-GTM achieves optimal performance compared with other state-of-the-art methods. More specifically, SR-GTM has improvements of 0.77%, 0.38%, and 3.63% over the best baseline method in P@20 and has improvements of 2.91%, 2.52%, and 2.49% in MRR@20, respectively.

Influence of data mining technology in information analysis of human resource management on macroscopic economic management.

The purposes are to manage human resource data better and explore the association between Human Resource Management (HRM), data mining, and economic management. An Ensemble Classifier-Decision Tree (EC-DT) algorithm is proposed based on the single decision tree algorithm to analyze HRM data. The involved single decision tree algorithms include C4.5, Random Tree, J48, and SimpleCart. Then, an HRM system is established based on the designed algorithm, and the evaluation management and talent recommendation modules are tested. Finally, the designed algorithm is compared and tested. Experimental results suggest that C4.5 provides the highest classification accuracy among the single decision tree algorithms, reaching 76.69%; in contrast, the designed EC-DT algorithm can provide a classification accuracy of 79.97%. The proposed EC-DT algorithm is compared with the Content-based Recommendation Method (CRM) and the Collaborative Filtering Recommendation Method (CFRM), revealing that its Data Mining Recommendation Method (DMRM) can provide the highest accuracy and recall, reaching 35.2% and 41.6%, respectively. Therefore, the data mining-based HRM system can promote and guide enterprises to develop according to quantitative evaluation results. The above results can provide a reference for studying HRM systems based on data mining technology.

A movie recommendation method based on users' positive and negative profiles

In the traditional content-based recommendation method, we usually use the movies users watched before or rated to represent their profile. However, there are many movies that users have never seen or rated. For an unrated movie, there are two possibilities: maybe the user likes it or does not like it. In this paper, we first focus on how to identify users' preferences for movies by using a collaborative filtering algorithm to predict the users’ movie ratings. We can then create two movie lists for each user, where one is the movies the user likes (with higher predicting or true ratings), and the other is the movies the user does not like (with lower predicting or true ratings). Based on these two movie lists, we establish a user positive profile and a user negative profile. Therefore, our algorithm will recommend to users movies that are most similar to their positive profile and most different from their negative profile. Finally, our experiments show that our method can improve the MAE index of the traditional collaborative filtering method by 12.54%, the MAPE index by 17.68%, and the F1 index by 10.16%.

사용자의 온오프 라인 쇼핑에서 상품 추천을 위한 전이학습 모델 비교

Recently, deep learning based services has been actively developed in various fields. Especially, in the online distribution environment some Applications with deep learning based on large amounts of data and user information began to be applied to real systems. In this paper, we develop an application system which recommends goods after learning and predicting using deep learning by presenting good image that user wants to purchase in online market sites. At this time, we apply three revised models based on the transfer learning model for automatic image recognition. After image recognition of goods, the system produces candidate goods and requires filter condition input for goods he wants to purchase. Finally, the system recommends goods by content-based recommendation method. In this paper, we experiment three revised models of image recognition which are used as core techniques in the system. These models are fully connected network, k-nearest neighbors search and linear SVM and they act as a classifier within the transfer learning model architecture. We experiment and compare performance evaluation for the revised models.

Music recommender using deep embedding-based features and behavior-based reinforcement learning

With the rapid increase of digital music on online music platforms, it has become difficult for users to find unknown but interesting songs. Although many collaborative filtering or content based recommendation methods have been proposed, they have various relatively serious some problems, including cold start, diversity of recommendations. etc. Therefore, we propose a reinforcement personal music recommendation system (RPMRS) to address these problems. RPMRS comprises two main components. First, deep representation of audio and lyrics extracted by WaveNet and Word2Vec models, respectively, and apply a proposed content based recommendation method from these. Second, we employ reinforcement learning is to learn user preferences from their song playing log. Experimental results confirm, that hybrid features are superior to audio or lyrics based features for content recommendation, largely because independent audio features significantly outperform lyrics features; and reinforcement learning improves personalized recommendations. Overall, the proposed RPMRS provides dynamic and personalized music recommendations for the user.

A vocabulary recommendation method for spatiotemporal data discovery based on Bayesian network and ontologies

ABSTRACTIn the research field of spatiotemporal data discovery, how to utilize the semantic characteristics of spatiotemporal datasets is an important topic. This paper presented a content-based recommendation method, and applied Bayesian networks and ontologies into the vocabulary recommendation process for spatiotemporal data discovery. The source data of this research was from the MUDROD (Mining and Utilizing Dataset Relevancy from Oceanographic Datasets) search platform. From the historical search log, major keywords were extracted and organized according to ontologies in a hierarchical structure. Using the search history, the posterior probability between each subclass and their super class in the ontologies was calculated, indicating a recommendation likelihood. We created a Bayesian network model for inference based on ontologies. This model can address the following two objectives: (1) Given one class in the ontology, the model can judge which class has the biggest likelihood to be selected for recommendation. 2) Based on the search history of a user, the Bayesian network model can judge which class has the biggest probability to be recommended. Comparison experimentation with existing system and evaluation experimentation with expert knowledge show that this method is specifically helpful for spatiotemporal data discovery.

Development and Quality Analysis of Decision Support Systems as Software for Scholarship Recommendation in Higher Education

This study aimed to develop a system that provides scholarship data, recommendations based on students’ ability and that ensures that the developed system provides appropriate recommendations according to the students’ ability by examining the results of system recommendations on academic, organizational, and student achievement. The research design used a Research and Development (R & D) method and procedures in the development of a scholarship recommendation system using the Extreme Programming (XP) development procedure with four stages: planning, design, coding, and testing. Testing of recommendations was performed by using a guide confusion matrix which tests seven aspects such as accuracy, recall, precision, prevalence, error rate, false positive rate, and specificity. The results of the research showed that: (1) the recommendation system for the selection of scholarships in universities resulted in recommendations using a content-based recommendation method with the weighting of the term frequency-inverse document frequency. (2) Testing the accuracy of the recommendation resulted in a 100% success rate.

A novel recommendation system based on semantics and context awareness

The existing content-based recommendation methods have two major limitations. First, due to the defects of the items and the user model matching algorithms, the recommendation results are very narrow. Second, scant attention is paid to the scenario, making the recommendation system not context-aware. It is essential to improve user satisfaction through high-quality recommendation. In this paper, two state-of-the-art methods are analyzed and extended to enhance recommendation performance. The first method is the context-aware recommender, which integrates context information into the recommendation process. The second method is the semantic analysis-based recommender, which incorporates domain semantics. Despite their compatibility, the challenge is to combine them in a way that will fully exploit their potential. An improved content-based model is proposed in this paper incorporating both semantics and context. Context-aware recommendation is performed to improve sensitivity to the context. Semantic relevance-based instance similarity is computed to address the problem of narrowness. The proposed recommendation system is evaluated using metrics (for instance, recall metric) and paralleled with the current methods grounded on the content. Results demonstrate the superiority of the proposed system in terms of accuracy.

Penerapan Metode Content-Based Filtering Pada Sistem Rekomendasi Kegiatan Ekstrakulikuler (Studi Kasus di Sekolah ABC)

Abstract — ABC School is an educational organization. They operate their routine activities without information systems, especially for extracurricular activies. System recommendation was made to help the school. It provides to recommend extracurricular which is more suitable with student interest. Primary data source is obtained by interview and observation with headmaster of ABC School. Primary data used for analysis and design sytem. The analysis use one of data mining technique, which is content based recommendation. The content based recommendation method that used for this research is Naive Bayes. The result of this research is a recommendation system to show propability the extracurricular fo each student with some chosen attributes. Keywords — attribute, information gain, Naive Bayes , recommendation system

An Improved Content-based Recommendation Method Through Collaborative Predictions and Fuzzy Similarity Measures

An Improved Content-based Recommendation Method Through Collaborative Predictions and Fuzzy Similarity Measures

Comparison of content-based music recommendation using different distance estimation methods

With rapid growth in the online music market, music recommendation has become an active research area. In most current approaches, content-based recommendation methods play an important role. Estimation of similarity between music content is the key to these approaches. A distance formula is used to calculate the music distance measure, and music recommendations are provided based on this measure. However, people have their own unique tastes in music. This paper proposes a method to calculate a personalized distance measure between different pieces of music based on user preferences. These methods utilize a randomized algorithm, a genetic algorithm, and genetic programming. The first two methods are based on Euclidean distance calculation, where the weight of each music feature in the distance calculation approximates user perception. The third method is not limited to Euclidean distance calculation. It generates a more complex distance function to estimate a user's music preferences. Experiments were conducted to compare the distance functions calculated by the three methods, and to compare and evaluate their performance in music recommendation.

User-centered profile representation for recommendations across multiple content domains

This paper proposes a content-based recommendation method tuned for multiple content domains. Most content-based recommender systems use content attributes (e.g. description of content and category of content) to represent items and user profiles. Th