Automatic topic discovery of online hospital reviews using an improved LDA with Variational Gibbs Sampling

Groof, Richard de; Xu, Haiping

doi:10.1109/bigdata.2017.8258417

Cited by 8 publications

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Next, we implemented LDA using a Gibbs Sampling approach. This approach is fast, efficient, and widely used in LDA applications (de Groof & Xu, 2017). We first set the necessary parameters for this approach and then ran the analysis.…”

Section: Latent Dirichlet Allocation (Lda)mentioning

confidence: 99%

The Interplay of Data-Driven Organizations and Data Spaces: Unlocking Capabilities for Transforming Organizations in the Era of Data Spaces

Hupperz,

Gieß

2024

Proceedings of the Annual Hawaii International Conference on System Sciences

View full text Add to dashboard Cite

Artificial intelligence (AI)-based systems, such as AI companions, have been increasingly used to meet the needs of individuals who experience loneliness. In this current study, we sought to identify the mechanism underlying human-AI interactions in the mental health context. We use a Latent Dirichlet Allocation (LDA) approach to analyze a sample of user-generated content consisting of rich data on AI companion app's reviews over a two-year period. We extracted five positive topics (i.e., perceived humanness, perceived emotional support, perceived AI's friendship, perceived (less) loneliness, and mental health benefits) and four negative topics (i.e., perceived lack of conscientiousness, perceived incredibility, perceived violation of privacy, and perceived creepiness of AI) from our analysis. Our AI-based emotional support model suggests that these positive and negative characteristics are interrelated. Our study provides an understanding of the relationship between AI companions and human users in light of research showing the effectiveness of an AI-based intervention for mental health care.

show abstract

Section: Latent Dirichlet Allocation (Lda)mentioning

confidence: 99%

The Interplay of Data-Driven Organizations and Data Spaces: Unlocking Capabilities for Transforming Organizations in the Era of Data Spaces

Hupperz,

Gieß

2024

Proceedings of the Annual Hawaii International Conference on System Sciences

View full text Add to dashboard Cite

show abstract

“…To validate the effectiveness of LDME, we conduct abundant experiments on Amazon.com dataset 1 and Yelp for RecSys. 2 Amazon and Yelp datasets are two public datasets.…”

Section: A Datasetsmentioning

confidence: 99%

“…Recently latent vector embedding has become a popular research spot, by its powerful ability to represent the complex relationships, where can be applied in various areas, such as text mining [1], POI prediction [2] and recommender system [3]. Basically, latent vector embedding model makes efforts to measure the non-linear relationships, by mapping existing items (words, POIs, or users and items) into a k-dimension latent space, then a non-linear relationship in original space can be transferred into a relatively linear relationship in this latent space.…”

Section: Introductionmentioning

confidence: 99%

Improving Recommendations by Embedding Multi-Entity Relationships With Latent Dual-Metric Learning

Yang

Han

et al. 2019

IEEE Access

View full text Add to dashboard Cite

Recently, latent vector embedding has become a research hotspot, with its great representative ability to measure the latent relationships among different views. However, most researches utilize the inner product of latent vectors as the representation of relationships, and they develop some embedding models based on this theory. In this paper, we take deep insight into the existing embedding models and find that utilizing the inner product may increase several problems: 1) in latent space, the inner product among three vectors may violate triangle principle; 2) the inner product cannot measure the relationships between vectors in the same category, such as user and user and item and item; and 3) the inner product cannot catch the collaborative relationships (user-user and item-item) for collaborative filtering. Along with this line, we propose a latent vector embedding model for collaborative filtering: latent dual metric embedding (LDME), which utilizes the dual-Euclidean distance in latent space, instead of the inner product, to represent different types of relationships (user-user, item-item, and user-item) with a uniform framework. Specifically, we design an embedding loss function in LDME, which can measure the close and remote relationships between entities, tackle the above problems, and achieve a more clear, well-explained embedding result. Extensive experiments are conducted on several real-world datasets (Amazon, Yelp, Taobao, and Jingdong), where the expiring results demonstrate that LDME can overperform some state-of-the-art user-item embedding models and can benefit the existing collaborative filtering models. INDEX TERMS Latent vector embedding, metric learning, collaborative filtering, recommender system.

show abstract

“…Session content was mapped to 80 relevant topics using a topic modeling paradigm that uses natural language processing and statistical algorithms to find themes or topics in a large corpus of documents [1]. This paradigm has successfully identified themes in several different domains, including scientific journal abstracts [2], reviews on Yelp [3] and bioinformatics [4]. This paper shows how a topic model can be used to analyze a large curriculum.…”

Section: Introductionmentioning

confidence: 99%

Using a Topic Model to Map and Analyze a Large Curriculum

Takizawa

2023

Preprint

View full text Add to dashboard Cite

A qualitative and quantitative understanding of curriculum content is critical for knowing whether it’s meeting its learning objectives. Curricula for medical education present challenges due to amount of content, the diversity of topics and the large number of contributing faculty. To create a manageable representation of the content in the pre-clerkship curriculum at Yale School of Medicine, a topic model was generated from all educational documents given to students during the pre-clerkship period. The model was used to quantitatively map content to school-wide competencies. The model measured how much of the curriculum addressed each topic and identified a new content area of interest, gender identity, whose coverage could be tracked over four years. The model also allowed quantitative measurement of integration of content within and between courses in the curriculum. The methods described here should be applicable to curricula in which texts can be extracted from materials.

show abstract

Automatic topic discovery of online hospital reviews using an improved LDA with Variational Gibbs Sampling

Cited by 8 publications

References 12 publications

The Interplay of Data-Driven Organizations and Data Spaces: Unlocking Capabilities for Transforming Organizations in the Era of Data Spaces

The Interplay of Data-Driven Organizations and Data Spaces: Unlocking Capabilities for Transforming Organizations in the Era of Data Spaces

Improving Recommendations by Embedding Multi-Entity Relationships With Latent Dual-Metric Learning

Using a Topic Model to Map and Analyze a Large Curriculum

Contact Info

Product

Resources

About