Multimodal Aspect-Based Sentiment Analysis (MABSA), as an emerging task in the field of sentiment analysis, has recently received widespread attention. Its aim is to combine relevant multimodal data determine polarity a given aspect text. Researchers have surveyed both aspect-based analysis and but, best our knowledge, there no survey on MABSA. Therefore, order assist related researchers know MABSA better, we research work recent years. Firstly, concepts were introduced. Secondly, existing...

Discovering valuable learning path patterns from learner online data can provide follow-up learners with effective reference and improve their experience effects. In this paper, a personalized recommendation method based on weak concept mining is proposed. Firstly, according to the degree of mastery historical learners, maps different types are generated by clustering association rule algorithms. A set paths then automatically through topological sorting algorithm. Secondly, long short-term...

Time series remote sensing image is an important resource for dynamic monitoring of resources and environment, its abundant time spectrum information can be used to characterize the change vegetation coverage. This paper proposes a comprehensive clustering pixel classification method extracting dynamics based on Landsat normalized difference index (NDVI). uses time-division algorithm fitting time-series NDVI firstly. And Markov random field optimized (MRF) semi-supervised warping (DTW)...

This article has been retracted. A retraction notice can be found at https://doi.org/10.3233/JIFS-219433.

A large volume of research documents are available online for us to access and analysis. It is very important detect mine the dynamics topics from these corpora. In this paper, we propose an improved method by introducing WordNet LDA. This approach find latent corpora, then many methods analyze those topics. We apply methodology two document collections: 1940 papers NIPS 00-13 (1987–2000) 2074 14-23 (2001–2010). Six experiments conducted on corpora experimental results show that our better...

Person–job fit, which aims to predict the matching degree between a resume and job, has become an effective way overcome information overload in recruitment market. Existing studies on person–job fit usually focus representation learning of textual data jobs resumes. is highly nonlinear complex problem that affected by several fields features. We assume it would bring benefits comprehensively consider numerical features, categorical features resumes jobs. To this end, we propose novel model...

Traditional topic detection method can not be applied to the microblog directly, because text is a kind of short, fractional and grass-roots text. In order detect hot in effectively, we propose based on combination latent semantic analysis structural property. According dialogic property microblog, our proposed firstly creates space replies thread, with aim solve data sparseness problem; secondly, create model analysis; finally, computation combined time information. We then adopt...

Text similarity measurements are the basis for measuring degree of matching between two or more texts. Traditional large-scale detection methods based on a digital fingerprint have advantage high speed, which only suitable accurate detection. We propose method Chinese text measurement feature phrase semantics. Natural language processing (NLP) technology is used to pre-process and extract keywords by Term frequency-Inverse document frequency (TF-IDF) model further screen out words. get exact...

With the increase in needs for personalized learning of online students, knowledge tracing (KT), a technique aimed at state student’s mastery and predicting performance future exercises, has become hot topic research. The behavioral features exhibited during students’ process bear information that impacts mastery. To study influence behaviors on process, we propose Precise Modeling Learning Process based Multiple Behavioral Features Knowledge Tracing model (MBFKT), which models by making use...

This paper proposed a reversible watermarking scheme for 2D vector drawings based on difference expansion DE. Few researches have been conducted method of till now. In our scheme, Information data including watermark is embedded into the nearest pairs vertexes sequence in x-direction or y-direction instead neighbouring vertexes, and as result capacity will be increased shape distortions reduced. The Hash value from original also part order to exactly estimate completeness extracted data....

Global gender disparity in science is an unsolved problem. Predicting has important role analysing the gap through online data. We study this problem within UK, Malaysia and China. enhance accuracy of existing prediction tools names that can predict sex Chinese characters English simultaneously with more precision. During our research, we found there no free forecasting tool to arbitrary number names. addressed shortcoming by providing a requests. demonstrate experimental results. show...

In this paper, we present a novel Local Sensitive Dual Concept Learning (LSDCL) method for the task of unsupervised feature selection. We first reconstruct original data matrix by proposed dual concept learning model, which inherits merit co-clustering based mechanism more interpretable and compact reconstruction. then adopt local sensitive loss function, emphasizes on most similar pairs with small errors to better characterize structure data. way, our can select features clustering results...

There have been many domain-specific keyword extraction researches, but micro-blog- oriented is just beginning. This paper researches into the from Chinese micro-blog. Taking characteristics of micro-blog account, such as short, topic divergence, etc., we propose a method based on combination multi features. Firstly create graph model co-occurrence between words, get kind weight created model. The sometimes same. In order to solve this problem, secondly proposes semantic space detection...

Teaching evaluation is an important part of the teaching process, and effective measure to improve method. In order automatically analyse massive texts on internet, this paper proposes apply sentiment classification technology analysis texts, introduces a method based dictionary. deal with problem frequent appearance new words in these propose recognise automatically, then use recognised expand Experimental results show that expanded dictionary increases recall rate system successfully,...

As one of the fundamental tasks in natural language processing, Multi-Label Text Classification (MLTC) is used to mark or more relevant labels for a given text from large set labels. Existing MLTC methods have increasingly focused on improving classification effectiveness by fusing correlations Still, research suffers difficulties comprehensively extracting features and distinguishing similar This paper proposed multi-label model based keyword extraction attention mechanism. The using...

In light of the current uneven distribution aerobics course online resources, this paper presents research on classification method teaching resources based artificial intelligence technology, constructs display education resource management system realizes information characteristics, and evaluation algorithm aerobiology resources. The achieves its design goal. Finally, experiment shows that intelligence-based for is highly practicable fully meets requirements.

Text similarity computing is the core issue that question-answering system needs to solve.It mainly used filter out existed problems which are similar user's questions from database.Because of low recall domain keywords in text based on traditional semantic dictionary, this paper proposed a short method field agriculture extended version <<Tongyicicilin>> referred as <<CiLin>>.This propose consider both and correlation when calculate words' final similarity.The experimental results show...

Text similarity computing is the core issue that question-answering system needs to solve. It mainly used filter out existed problems which are similar users questions from database. Because of low recall domain keywords in text based on traditional semantic dictionary, this paper proposed a short method field agriculture extended version &lt;&lt;Tongyicicilin&gt;&gt; referred as &lt;&lt;CiLin&gt;&gt;. This propose consider both and correlation when calculate words final similarity. The...

At present, the Chinese text field is facing challenges from low resource issues such as data scarcity and annotation difficulties. Moreover, in domain of cigarette tasting, tasting texts tend to be colloquial, making it difficult obtain valuable high-quality texts. Therefore, this paper, we construct a dataset (CT2023) propose novel classification method based on ERNIE Comparative Learning for Low-Resource scenarios (ECLLR). Firstly, address limited vocabulary diversity sparse features...