Time-evolving Text Classification with Deep Neural Networks

He, Yu; Li, Jianxin; Song, Yangqiu; He, Mutian; Peng, Hao

doi:10.24963/ijcai.2018/310

Cited by 38 publications

(31 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the recent remarkable achievements of deep learning methods, several methods have been proposed to investigate the problem of concept drift using deep learning methods. For example, [107,152] proposed an RNN model for dealing with the problem of the concept drift for time series anomaly detection to address challenges posed by sudden or regular changes in normal behavior. The model is trained incrementally as new data becomes available and is capable of adapting to the changes in the data distribution.…”

Section: Recommendation and Future Research Directionmentioning

confidence: 99%

Recommender System Based on Temporal Models: A Systematic Review

et al. 2020

View full text Add to dashboard Cite

Over the years, the recommender systems (RS) have witnessed an increasing growth for its enormous benefits in supporting users’ needs through mapping the available products to users based on their observed interests towards items. In this setting, however, more users, items and rating data are being constantly added to the system, causing several shifts in the underlying relationship between users and items to be recommended, a problem known as concept drift or sometimes called temporal dynamics in RS. Although the traditional techniques of RS have attained significant success in providing recommendations, they are insufficient in providing accurate recommendations due to concept drift problems. These issues have triggered a lot of researches on the development of dynamic recommender systems (DRSs) which is focused on the design of temporal models that will account for concept drifts and ensure more accurate recommendations. However, in spite of the several research efforts on the DRSs, only a few secondary studies were carried out in this field. Therefore, this study aims to provide a systematic literature review (SLR) of the DRSs models that can guide researchers and practitioners to better understand the issues and challenges in the field. To achieve the aim of this study, 87 papers were selected for the review out of 875 total papers retrieved between 2010 and 2019, after carefully applying the inclusion/exclusion and the quality assessment criteria. The results of the study show that concept drift is mostly applied in the multimedia domain, then followed by the e-commerce domain. Also, the results showed that time-dependent neighborhood models are the popularly used temporal models for DRS followed by the Time-dependent Matrix Factorization (TMF) and time-aware factors models, specifically Tensor models, respectively. In terms of evaluation strategy, offline metrics such as precision and recalls are the most commonly used approaches to evaluate the performance of DRS.

show abstract

Section: Recommendation and Future Research Directionmentioning

confidence: 99%

Recommender System Based on Temporal Models: A Systematic Review

et al. 2020

View full text Add to dashboard Cite

show abstract

“…We use the model from the epoch when the model achieves the best result on the development set for the final model. He et al (2018) propose an evolving framework to train document classifiers. We re-implement two classifiers, RCNN and HAN with diachronic propagation learning strategy, which achieved the best performances in their paper.…”

Section: Dannmentioning

confidence: 99%

Neural Temporality Adaptation for Document Classification: Diachronic Word Embeddings and Domain Adaptation Models

Huang¹,

Paul²

2019

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics

View full text Add to dashboard Cite

Language usage can change across periods of time, but document classifiers models are usually trained and tested on corpora spanning multiple years without considering temporal variations. This paper describes two complementary ways to adapt classifiers to shifts across time. First, we show that diachronic word embeddings, which were originally developed to study language change, can also improve document classification, and we show a simple method for constructing this type of embedding. Second, we propose a time-driven neural classification model inspired by methods for domain adaptation. Experiments on six corpora show how these methods can make classifiers more robust over time.

show abstract

“…The most widely used type of recommendation algorithm is based on collaborative filtering. Such algorithms can automatically learn the hidden features of users and items, which can help identify potential interests of users [19] [20]. However, collaborative filtering algorithms present data sparseness and cold start problems.…”

Section: Introductionmentioning

confidence: 99%

Score Prediction Algorithm Combining Deep Learning and Matrix Factorization in Sensor Cloud Systems

Gong

et al. 2021

IEEE Access

View full text Add to dashboard Cite

In this era of exponential growth in the scale of data, information overload has become an urgent problem, and the use of increasingly flexible sensor cloud systems (SCS) for data collection has become a mainstream trend. Recommendation algorithms can search massive data sets to uncover information that meets the needs of users based on their interests. To improve the accuracy of recommendation scoring, this paper proposes a score prediction algorithm that combines deep learning and matrix factorization. To address the problem of sparse scoring data, our study employs a sensor cloud system to collect data information, preprocesses the collected information, and then uses a deep learning model combined with explicit and implicit feedback to generate recommendations. The proposed algorithm, MF-NeuRec, combines fusion matrix decomposition and the NeuRec model score prediction algorithm. The algorithm employs user-based and item-based NeuRec algorithms to extract the feature vectors of users and items under implicit feedback data. The obtained user and item feature vectors are integrated in a certain ratio through the use of matrix decomposition under the display feedback data. The user and item feature vectors obtained by the algorithm are merged and analyzed to predict how users will rate items. Experiments demonstrate that the algorithm can improve the accuracy of recommendations.

show abstract

Time-evolving Text Classification with Deep Neural Networks

Cited by 38 publications

References 0 publications

Recommender System Based on Temporal Models: A Systematic Review

Recommender System Based on Temporal Models: A Systematic Review

Neural Temporality Adaptation for Document Classification: Diachronic Word Embeddings and Domain Adaptation Models

Score Prediction Algorithm Combining Deep Learning and Matrix Factorization in Sensor Cloud Systems

Contact Info

Product

Resources

About