A Word Embedding Model for Topic Recommendation

Kannan, Megala S; Mahalakshmi, G. S.; Smitha, E S; Sendhilkumar, S.

doi:10.1109/icicct.2018.8473055

Cited by 1 publication

(1 citation statement)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Perplexity is seen as a good measure of performance for LDA. Based on the research of Kannan, Mahalakshmi, Smitha, and Sendhilkumar (2018), The idea is that you keep a holdout sample, train your LDA on the rest of the data, then calculate the perplexity of the holdout. The perplexity could be given by the formula:…”

Section: Perplexitymentioning

confidence: 99%

Province of Laguna Legislative Management and Tracking System with the Application of Latent Dirichlet Allocation (LDA) Algorithm

Cruz¹,

Balahadia²

2023

IJCSR

View full text Add to dashboard Cite

Purpose -The legislative branch of a province is generally in charge of making laws. In addition to this, they are also in charge of enacting programs and policies for the general well-being of the citizens within the province. The general public may not be able to keep track of legislative performances since there is a growing number of legislative-generated documents. This study developed a system that integrated the application of topic modeling in crafting ordinances, resolutions, and policies for the Province of Laguna.Method -This research employed the SCRUM methodology process in the development of the system and Latent Dirichlet Allocation (LDA) topic modeling using R language to classify text within a document to a particular topic and created model through Observation-based evaluation and Quantitative metrics such as Perplexity and Coherence to determine k-value (number of topics) based on the corpus wherein it was collected in the Twitter and Legislative Management System Portal.Results -The results showed the LDA used returned the optimal value for perplexity and coherence which was determined by testing different k-values ranging from 1 to 2 which is presented to users as a line graph. The developed system provided a system module 1163 that can enable users to find the optimal number of topics (k value) and present the results in a visually appealing interface on the user's account portal which gives insights into what possible new ideas in formulation ordinances, resolutions, and policies in Laguna. Conclusion -The developed system in this study allows legislators of the province of Laguna to collect public posts from the social networking site Twitter and use Latent Dirichlet Allocation (LDA) topic modeling. It also provides an interactive graph that allows users to explore the LDA model generated by the system and helps to reveal topics of concern from the community that leads to government officials in formulating policies and ordinances appropriate for the needs of the community.Recommendations -It is recommended to develop an additional module that automatically generates the topic model based on the selected LDA evaluation procedure and should be tested in a larger-sized corpus to further test its capabilities as well as to improve the list of the stop words and noise removal feature. Implications -The system can be used to simply accomplish the document trail page where users can preview document details and the application of the visualization techniques in the system helps to facilitate the to provide an impression by extracting words, and topics that can be a basis of crafting programs and priorities of the government officials in taking actions to the citizen concerns.

show abstract

Section: Perplexitymentioning

confidence: 99%