Exploring Scientists’ Research Behaviors Based on LDA

Li, Benji; Gu, Weiwei; Lu, Yahan; Wu, Chensheng; Chen, Qinghua

doi:10.1007/978-3-030-22868-2_14

Cited by 1 publication

(2 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The objective is to provide brief descriptions of the members of a cluster. As a result, large clusters can be processed efficiently, and valuable statistical relationships can be generated for various tasks, including clustering, summarizing, evaluating, and determining similarity between clusters (Li et al, 2019).…”

Section: Latent Dirichlet Allocation (Lda) Algorithmmentioning

confidence: 99%

See 1 more Smart Citation

A predictive analytics solution matching job seekers' talent and employers' demands based on machine learning

Rahmani,

Groot,

Rahmani

2023

Preprint

View full text Add to dashboard Cite

The Covid-19 Pandemic has led to an increase in online recruitment. Many job seekers and employers post information about their talents, characteristics, and demands on job search websites where a large amount of data is cumulated in an unparsed way. Therefore, job seekers and employers cannot always find their way through this. This research used a sample of data from the most dependable employment and job search website in Iran. By examining the database provided by the website, the study aims to identify, cluster, predict and analyze data related to job seekers and the labour market. This study utilized the CRISP-DM methodology to analyze job seeker and employers' demands data, utilizing clustering algorithms such as self-organizing maps (SOM), Fuzzy C-means (FCM), K-Means, and latent Dirichlet allocation (LDA), with validation using the silhouette criterion. Then Multi-layer Perceptron (MLP), Naive Bayes, Learning Vector Quantization (LVQ), Support Vector Machine (SVM), and Language Model (LM) classification algorithms were used for prediction and classification processes. Results showed that the SOM algorithm had the best clustering with an average silhouette index of 0.92 for job seekers and 0.89 for the labour market. The MLP algorithm with SOM clustering was chosen as the best algorithm for prediction and classification processes with an accuracy of 0.83 which is a relatively high level of accuracy in predictive analytics. JEL code: J23, J21, J24, C38, C55.

show abstract

Section: Latent Dirichlet Allocation (Lda) Algorithmmentioning

confidence: 99%

“…The fuzzy clustering method was developed by Dunn in 1974 and improved by Bezdek in 1981 when he applied fuzzy theory to the clustering process (Y. Li, Chen, & Li, 2019). An important aspect of FCM is using the Euclidean distance, which is based on a criterion of the distance between samples.…”

Section: 83fuzzy C-means Algorithmmentioning

confidence: 99%