Protobi

← All help posts

View post

Protobi can give automated sentiment analysis scores for text verbatims using leading AI libraries from [Indico](https://indico.io/), OHMSense and [ParallelDots](https://paralleldots.com/).

Above is a stock photo of classic candy hearts, annotated with the sentiment analysis scores from Protobi. We imagine you’d use this to evaluate open-end survey responses. But the candy hearts illustrate the strengths and limitations of sentiment analysis.

## How to access

To use sentiment analysis on a question, first select the element. Then, from the "Advanced" button on the toolbar choose "Sentiment (BETA)."

From the Sentiment dialog choose which AI library you want to use.

## How it works

These AI libraries score text on a scale from 0 to 100%, where 100% is very positive, 0% is very negative, and 50% is neutral. Here’s one example scoring a random list of adjectives:

At heart, computerized sentiment analysis is a bit simplistic from a human perspective. The computer doesn’t really get in the mind of the author to divine the actual sentiment. It scores the words within the text and returns an aggregate summary score.

## Strengths

Sentiment analysis is reasonably good at appropriately scoring words that are widely considered positive or negative.

**“True love” (98%)
“Best day” (97%)
“Laugh”: (90%)**

**“Fart monster” (2%)**

And it rates the following as having a low sentiment even though might intend to communicate quite the opposite:

And it m

In this sense, automated sentiment analysis is effective at quickly sorting through lots of verbal expressions and extracting general trends.

## Limitations

The scores in the candy hearts image are taken directly from the algorithm. We assume it gives the following ratings because the computer simply doesn’t understand the experience:

**“XOXO” (67%) 
“First kiss” (47%)**

The following phrase has a high sentiment score even though the actual meaning may not be all that positive:

**“You’re really nice…” (98%)**

And it rates the following as having a low sentiment even though the intent may be to communicate quite the opposite:

**“You’re not half bad” (32%)**

It may even completely miss subtle British-vs-American interpretations, at least according to [this Guardian reporter](https://www.youtube.com/watch?v=VWrfaNN7DGg "Go to YouTube video") who explains that in British English, this is not a high compliment:

**“Quite good” (98%)
**

Publishing

Date

Status

Slug edit

Content

Thumbnail

Categories Manage

Tools

AI Database

Publishing

Content

Tools

Danger zone