11. Social Media Analytics for HC Flashcards
how is social media used in hc
syndromic surveillance systems e.g. HealthMap use global media sources to provide information on disease activity around the world
what are the 2 different types of text mining
data mining: exploring textual data patterns e.g. latent topics, outliers
NLP processing: making inferences based on partial understanding of NLP text e.g. information extraction & QA
is data mining deep or shallow
shallow
is NLP processing deep or shallow
deep
what are direct applications of text mining
using data mining to answer specific questions
using data to derive insights and patterns
what are indirect applications of text mining
assist information access & organisation
what are some text mining methods
- data mining style
- information retrieval style
- NLP style
- ML style
what is the data mining style of text mining
frequent pattern finding, outlier detection & association analysis
what is the information retrieval style of text mining
extracting topics, using weight terms & text similarity measures
what is the NLP style of text mining
question answering, relation extraction, entity extraction, sentiment analysis
what is the ML style of text mining
unsupervised/semi-supervised ML via mixture models & dimensionality reduction
what is a mixture model
a model that uses a combination of multiple probability distributions, where the each distributions output is weighted such that all outputs from the distributions sum to 1
what is LDA
latent dirichlet allocation
graphical model is designed to examine a corpus of documents and form clusters of semantically related words to summarise the topics of documents in the collection
is LDA a graphical model
yes
how does LDA work
