Dis-Cover AI Minds to Preserve Human Knowledge

Ranaldi, Leonardo; Fallucchi, Francesca; Zanzotto, Fabio Massimo

doi:10.3390/fi14010010

Cited by 16 publications

(7 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another method, called Knowledge Inference or Latent Knowledge Estimation, is associated with using EDM techniques to "assess the skills of students based on their responses to a problem-solving exercise" (p. 184) [21]. For example, natural language processing technologies (e.g., Transformers) have the potential to be used to measure human language knowledge acquisition [23]. This research sits the classification method, one of the commonly seen methods in predictive analysis [20].…”

Section: Learning Analytics and Educational Data Miningmentioning

confidence: 99%

Predicting Student Performance Using Clickstream Data and Machine Learning

Liu

Fan

et al. 2022

Education Sciences

View full text Add to dashboard Cite

Student performance predictive analysis has played a vital role in education in recent years. It allows for the understanding students’ learning behaviours, the identification of at-risk students, and the development of insights into teaching and learning improvement. Recently, many researchers have used data collected from Learning Management Systems to predict student performance. This study investigates the potential of clickstream data for this purpose. A total of 5341 sample students and their click behaviour data from the OULAD (Open University Learning Analytics Dataset) are used. The raw clickstream data are transformed, integrating the time and activity dimensions of students’ click actions. Two feature sets are extracted, indicating the number of clicks on 12 learning sites based on weekly and monthly time intervals. For both feature sets, the experiments are performed to compare deep learning algorithms (including LSTM and 1D-CNN) with traditional machine learning approaches. It is found that the LSTM algorithm outperformed other approaches on a range of evaluation metrics, with up to 90.25% accuracy. Four out of twelve learning sites (content, subpage, homepage, quiz) are identified as critical in influencing student performance in the course. The insights from these critical learning sites can inform the design of future courses and teaching interventions to support at-risk students.

show abstract

Section: Learning Analytics and Educational Data Miningmentioning

confidence: 99%

Predicting Student Performance Using Clickstream Data and Machine Learning

Liu

Fan

et al. 2022

Education Sciences

View full text Add to dashboard Cite

show abstract

“…However, as a transformer-based model, BERT relies heavily on the knowledge derived from the experience learned from a large corpus [22]. In addition, BERT models require considerable computing resources in the training stage.…”

Section: Transformer-based Modelmentioning

confidence: 99%

“…In addition, BERT models require considerable computing resources in the training stage. BERT appears to contain appropriate mechanisms for learning universal linguistic representations that are taskindependent [22]. Therefore, publicizing BERT models that have been pretrained based on various types of corpora can provide pretrained models to help other researchers finetune their models according to specific tasks, contributing to efficiency improvement of research and business practice.…”

Section: Transformer-based Modelmentioning

confidence: 99%

Text Mining of User-Generated Content (UGC) for Business Applications in E-Commerce: A Systematic Review

Liu

Zhang

et al. 2022

Mathematics

View full text Add to dashboard Cite

In the Web2.0 era, user-generated content (UGC) provides a valuable source of data to aid in understanding consumers and driving intelligent business. Text mining techniques, such as semantic analysis and sentiment analysis, help to extract meaningful information embedded in UGC. However, research on text mining of UGC for e-commerce business applications involves interdisciplinary knowledge, and few studies have systematically summarized the research framework and application directions of related research in this field. First, based on e-commerce practice, in this study, we derive a general framework to summarize the mainstream research in this field. Second, widely used text mining techniques are introduced, including semantic and sentiment analysis. Furthermore, we analyze the development status of semantic analysis in terms of text representation and semantic understanding. Then, the definition, development, and technical classification of sentiment analysis techniques are introduced. Third, we discuss mainstream directions of text mining for business applications, ranging from high-quality UGC detection and consumer profiling, to product enhancement and marketing. Finally, research gaps with respect to these efforts are emphasized, and suggestions are provided for future work. We also provide prospective directions for future research.

show abstract

“…USEs such as BERT (Devlin et al, 2019) are encoding semantic features in hidden layers (Jawahar et al, 2019;Miaschi et al, 2020). However, USEs' success in downstream tasks may be due to superficial heuristics (as supposed in McCoy et al (2019) and Ranaldi et al (2022)) and not to deep modeling of semantic features. Therefore, our study can contribute to this debate.…”

Section: Background and Related Workmentioning

confidence: 99%

Lacking the Embedding of a Word? Look it up into a Traditional Dictionary

Ruzzetti¹,

Ranaldi²,

Mastromattei³

et al. 2022

Findings of the Association for Computational Linguistics: ACL 2022

Self Cite

View full text Add to dashboard Cite

Word embeddings are powerful dictionaries, which may easily capture language variations. However, these dictionaries fail to give sense to rare words, which are surprisingly often covered by traditional dictionaries. In this paper, we propose to use definitions retrieved in traditional dictionaries to produce word embeddings for rare words. For this purpose, we introduce two methods: Definition Neural Network (DefiNNet) and Define BERT (DefBERT). In our experiments, DefiNNet and DefBERT significantly outperform state-of-the-art as well as baseline methods devised for producing embeddings of unknown words. In fact, DefiNNet significantly outperforms FastText, which implements a method for the same task-based on n-grams, and DefBERT significantly outperforms the BERT method for OOV words. Then, definitions in traditional dictionaries are useful to build word embeddings for rare words.

show abstract

Dis-Cover AI Minds to Preserve Human Knowledge

Cited by 16 publications

References 57 publications

Predicting Student Performance Using Clickstream Data and Machine Learning

Predicting Student Performance Using Clickstream Data and Machine Learning

Text Mining of User-Generated Content (UGC) for Business Applications in E-Commerce: A Systematic Review

Lacking the Embedding of a Word? Look it up into a Traditional Dictionary

Contact Info

Product

Resources

About