data science for studying social attitudes, latent cognitive factors
click to edit
topic modeling algorithm
inputs and configuration options
word2vec
hypothesis
golas
modeling tone and sentiment of news
modeling and traccking specific semantic topic
a case study: USA election of 2016 .
observation:analysis of share of voice
click to edit
click to edit
polls and news
click to edit
why to use it? modeling correlation of sales and advertiesement
the project for excelence in journalism
general sentiment
which fetures and sentiments are of interest
click to edit
to model tone of news coverage
click to edit
click to edit
tone of news models
what latent factors?
observed that: despite TR's statements had negative tone, but since his statements were unexpected, he made large the share of voice. being on the news headline made him stronger
components of the model
language model component
click to edit
ACT-GMM component
click to edit
induces a connection between linguistic categories that represent the names of entities and stable, affective meanings of these entities
click to edit
a particular measurement system for behaviors, identities, settings and modifiers, which I will collectively refer to as entities
click to edit
click to edit
click to edit
fundamental affective meaning of each entity
activity
potency
evaluation
evaluation dimension describes the goodness/badness of an entity
describes the powerfulness/weakness of a given entity
level of activity/passivity