Hierarchical attention model for personalized tag recommendation

Sun, Jun; Zhu, Mingyue; Jiang, Yuanchun; Liu, Yezheng; Wu, Le

doi:10.1002/asi.24400

Cited by 19 publications

(12 citation statements)

References 38 publications

(47 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…After years of application, the teaching platform stores a large amount of teaching materials and learning resources and uses educational data mining to analyze learners' learning behaviors, knowledge levels, and learning preferences in the learning process and to discover the defects and risks in the learning process, as shown in Figure 1. In summary, in order to solve the problems of blindness and inappropriate resources in the search for learning resources, the data recorded during the learning process are used rationally to discover learning resources that match students' own personality characteristics [16]. In this paper, we propose an individual learning resource recommendation model for students and study cross use of multidimensional data.…”

Section: Introductionmentioning

confidence: 99%

[Retracted] Personalized Recommendation Model of High‐Quality Education Resources for College Students Based on Data Mining

Fang

Qiu-yun

2021

Complexity

View full text Add to dashboard Cite

With the rapid development of information technology and data science, as well as the innovative concept of “Internet+” education, personalized e-learning has received widespread attention in school education and family education. The development of education informatization has led to a rapid increase in the number of online learning users and an explosion in the number of learning resources, which makes learners face the dilemma of “information overload” and “learning lost” in the learning process. In the personalized learning resource recommendation system, the most critical thing is the construction of the learner model. Currently, most learner models generally have a lack of scientific focus that they have a single method of obtaining dimensions, feature attributes, and low computational complexity. These problems may lead to disagreement between the learner’s learning ability and the difficulty of the recommended learning resources and may lead to the cognitive overload or disorientation of learners in the learning process. The purpose of this paper is to construct a learner model to support the above problems and to strongly support individual learning resources recommendation by learning the resource model which effectively reduces the problem of cold start and sparsity in the recommended process. In this paper, we analyze the behavioral data of learners in the learning process and extract three features of learner’s cognitive ability, knowledge level, and preference for learning of learner model analysis. Among them, the preference model of the learner is constructed using the ontology, and the semantic relation between the knowledge is better understood, and the interest of the student learning is discovered.

show abstract

Section: Introductionmentioning

confidence: 99%

[Retracted] Personalized Recommendation Model of High‐Quality Education Resources for College Students Based on Data Mining

Fang

Qiu-yun

2021

Complexity

View full text Add to dashboard Cite

show abstract

“…• CTR-SR [13] It is a hybrid method that combines the item-tag matrix, item content information, and item social information into a unified framework through a hierarchical Bayesian model. • HAM-TR [24] It models two important attentive aspects with a hierarchical attention model, which exploits two levels of attention to effectively aggregate different elements and different information of content information and collaborative information respectively.…”

Section: Baseline Methodsmentioning

confidence: 99%

“…Zhang et al [23] proposed an optimization model to integrate item contents into user interests where different impacts of item features on user preference toward an item have been extracted for item recommendation. Sun et al [24] proposed a hierarchical attention model, where collaborative embeddings and content embeddings were fused through an attention module. Wang et al [12] proposed a CTR model to recommend articles to users, which combined a collaborative filtering matrix decomposition algorithm based on hidden factors and a content analysis algorithm based on probabilistic topic models.…”

Section: Tag Recommendationmentioning

confidence: 99%

“…We use Recall@N, NDCG@N, and MRR@N as our evaluation metrics as in [24], where we do not choose Precision@N for the reason that an unobserved tag for an item may be since the tag is not suitable for the item, or that the tag is not considered by users to annotate to the item [12]. Recall@N calculates the hit ratio of the tested models:…”

Section: Evaluation Protocolsmentioning

confidence: 99%

See 1 more Smart Citation

Multi-Auxiliary Augmented Collaborative Variational Auto-encoder for Tag Recommendation

Yi¹,

Ren²,

Chen³

2022

Preprint

View full text Add to dashboard Cite

Recommending appropriate tags to items can facilitate content organization, retrieval, consumption and other applications, where hybrid tag recommender systems have been utilized to integrate collaborative information and content information for better recommendations. In this paper, we propose a multi-auxiliary augmented collaborative variational auto-encoder (MA-CVAE) for tag recommendation, which couples item collaborative information and item multi-auxiliary information, i.e., content and social graph, by defining a generative process. Specifically, the model learns deep latent embeddings from different item auxiliary information using variational auto-encoders (VAE), which could form a generative distribution over each auxiliary information by introducing a latent variable parameterized by deep neural network. Moreover, to recommend tags for new items, item multi-auxiliary latent embeddings are utilized as a surrogate through the item decoder for predicting recommendation probabilities of each tag, where reconstruction losses are added in the training phase to constrict the generation for feedback predictions via different auxiliary embeddings. In addition, an inductive variational graph auto-encoder is designed where new item nodes could be inferred in the test phase, such that item social embeddings could be exploited for new items. Extensive experiments on MovieLens and citeulike datasets demonstrate the effectiveness of our method.

show abstract

“…Boughareb et al show a new approach to recommend tags for scientific papers, which defines the relatedness between the tags attributed by users and the concepts extracted from the available sections of scientific papers based on statistical, structural and semantic aspects [59]. Sun et al presented a hierarchical attention model for personalized tag recommendation [60]. Lei et al introduce a tag recommendation by text classification that uses the capsule network with dynamic routing for tag recommendation.…”

Section: Subject Label Recommendationmentioning

confidence: 99%

Recommendations for Item Set Completion: On the Semantics of Item Co-Occurrence With Data Sparsity, Input Size, and Input Modalities

Vagliano¹,

Galke²,

Scherp³

2021

Preprint

View full text Add to dashboard Cite

We address the problem of recommending relevant items to a user in order to "complete" a partial set of items already known. We consider the two scenarios of citation and subject label recommendation, which resemble different semantics of item co-occurrence: relatedness for co-citations and diversity for subject labels. We assess the influence of the completeness of an already known partial item set on the recommender performance. We also investigate data sparsity through a pruning parameter, and the influence of using additional metadata. As recommender models, we focus on different autoencoders, which are particularly suited for reconstructing missing items in a set. We extend autoencoders to exploit a multi-modal input of text and structured data. Our experiments on six real-world datasets show that supplying the partial item set as input is helpful when item co-occurrence resembles relatedness, while metadata are effective when co-occurrence implies diversity. This outcome means that the semantics of item co-occurrence is an important factor. The simple item co-occurrence model is a strong baseline for citation recommendation. However, autoencoders have the advantage to enable exploiting additional metadata besides the partial item set as input, and achieve comparable performance. For the subject label recommendation task, the title is the most important attribute. Adding more input modalities sometimes even harms the result. In conclusion, it is crucial to consider the semantics of the item co-occurrence for the choice of an appropriate recommendation model and carefully decide which metadata to exploit.

show abstract

Hierarchical attention model for personalized tag recommendation

Cited by 19 publications

References 38 publications

[Retracted] Personalized Recommendation Model of High‐Quality Education Resources for College Students Based on Data Mining

[Retracted] Personalized Recommendation Model of High‐Quality Education Resources for College Students Based on Data Mining

Multi-Auxiliary Augmented Collaborative Variational Auto-encoder for Tag Recommendation

Recommendations for Item Set Completion: On the Semantics of Item Co-Occurrence With Data Sparsity, Input Size, and Input Modalities

Contact Info

Product

Resources

About