Our early stage researcher Shaheen Syed published and presented his latest paper entitled “Selecting Priors for Latent Dirichlet Allocation” at the IEEE International Conference on Semantic Computing (ICSC 2018) held in Laguna Hills, California, USA (Jan 31 – Feb 2, 2018).
His work explores symmetrical and asymmetrical Dirichlet priors when utilizing the topic model latent Dirichlet allocation (LDA). Since LDA is a Bayesian topic model, prior knowledge for words in topics, and topics in documents can be incorporated and his paper shows what the practical implications on the quality of topics can be when choosing one prior over the other.
The paper can be downloaded from here:
Selecting Priors for Latent Dirichlet Allocation
Syed, S., & Spruit, M. (2018). Selecting Priors for Latent Dirichlet Allocation. In The 12th IEEE International Conference on Semantic Computing (pp. 194–202). Laguna Hills, CA, USA: IEEE. http://doi.org/10.1109/ICSC.2018.00035