lda.qmd

# Topic Modeling {#sec-lda}

```{r setup}
#| echo: false
#| include: false

source("_common.R")
library(quanteda)

hippocorpus_corp <- read_csv("data/hippocorpus-u20220112/hcV3-stories.csv") |> 
  select(AssignmentId, story, memType, summary, WorkerId, 
         annotatorGender, openness, timeSinceEvent) |> 
  corpus(docid_field = "AssignmentId", 
         text_field = "story")

hippocorpus_dfm <- hippocorpus_corp |> 
  tokens(remove_punct = TRUE) |> 
  dfm() |> 
  dfm_remove("~")
```

::: callout-important
## This page is still under construction. Come back soon!
:::

**Topic Modeling**

LDA [@blei_etal_2003]

Dirichlet is [generallly pronounced either "Deereekleh" or "Deerishleh"](https://german.stackexchange.com/questions/48498/pronunciation-of-dirichlet)

@poldrack_etal_2012

```{r}
#| eval: false
library(seededlda)

lda <- textmodel_lda(dfm, k = 10, verbose = TRUE)
```

For larger corpora, set `batch_size` lower

https://psycnet.apa.org/record/2021-27454-001

## Supervised LDA {#sec-slda}

@blei_mcauliffe_2010

[sLDA in R](https://books.psychstat.org/textmining/topic-models.html#supervised-topic-modeling)

## Semi-Supervised LDA {#sec-seededlda}

[seededLDA in R](https://koheiw.github.io/seededlda/)

**An Example of Semi-Supervised LDA in Research:** @curini_valerio_2021

## BERTopic: Neural Topic Modeling

@grootendorst_2022

```{r}
#| eval: false
devtools::install_github("abresler/bertopic")
```


::: {.callout-tip icon="false"}
## Advantages of Topic Modeling

-   **:**
:::

::: {.callout-important icon="false"}
## Disadvantages of Topic Modeling

-   **:**
:::