presentation.qmd

---
title: "Text Analysis and NLP Techniques in Practice"
author: "Manu Alcalá Kovalski and Judah Axelrod"
title-slide-attributes:
  data-background-color: "#1696d2"
format: 
  revealjs:
    logo: www/images/urban-institute-logo.png
    incremental: true
editor: visual
output:
  html:
    number_sections: FALSE
    self_contained: TRUE
    code_folding: hide
    toc: TRUE
    toc_float: TRUE
    mathjax: null
    df_print: paged    
    css: !expr here::here("www", "web_report.css") 
editor_options:
  chunk_output_type: console
---



## Locate keywords

::: columns
::: {.column .incremental width="40%"}
-   Parse corpus of documents
-   Create dictionary of keywords
-   Find keywords across entire corpus or within a context
:::

::: {.column width="50%"}
![](images/DOL-eap.png){fig-align="center"}
:::
:::

------------------------------------------------------------------------

## Dictionary

![](images/communities-dictionary.png){.r-stretch width="90%"}

------------------------------------------------------------------------

## Global Search

![](images/communities-frequency-global-eap.png)

------------------------------------------------------------------------

## Window Search

![Searched for vulnerable communities within 20 words of mentions of
underserved, underrepresented, overburdened, vulnerable, marginalized,
disadvantaged, or communities of
concern.](images/communities-frequency-window-eap.png)

------------------------------------------------------------------------

## Topic Modeling

![Figure source: Blei, D.M (2012). Probabilistic topic models.
*Communication of the ACM 55*(4),
77-84.](images/topic-models-blei-2012.png)

------------------------------------------------------------------------

## Topics for AP News articles

![](images/ap-top-terms.png){width="100%" fig-align="center"}

------------------------------------------------------------------------

## Analyzing Twitter Trends

![](images/biden-tweet-freqs.png)

------------------------------------------------------------------------

## Sentiment Analysis

![](images/sentiment-analysis-trend.png){width="80%" fig-align="center"}