Bots to Combat Casual Sexism at Work

Representation of Women in Tech

National Center for Women & Information Technology (NCWIT), 25% of the computing workforce was female in 2015.
Hiring Across Technical roles: It’s an imbalance that’s reflected in company diversity reports. Microsoft, for example, boasts 29% female workers across its staff, but in technical positions only 17% are women. Of Google’s senior management and executive officer team, 17 are male while only three are women. Men make up 83% of Google’s engineering staff; Apple’s technical team is 80% male

Four in ten women in the United States have faced gender discrimination in their workplace. Sexism at the workplace is a global problem.
Casual Sexism-What is that?
But the masculine skew allows casual sexism to go unchecked, exemplified by ill-advised comments from certain chief executives. It’s so subtle that you’re often left wondering if it really happened, and when you try to talk about it, you sound like a crazy person because it’s so small – but it’s death by a thousand paper cuts

Global Level:

Personal Level:

Capturing Conversations- 24 percent of conversations in workplaces are written communications. Hence can be analyzed and monitored
Word Embeddings: Word Embeddings are Word converted into numbers

Humans can deal with text format quite intuitively whereas computers cannot
A computer can match two strings and tell you whether they are same or not. How do you make a computer understand that “Apple” in “Apple is a tasty fruit” is a fruit that can be eaten and not a company?
Conversion of the sentences to a dictionary

Frequency-based word embeddings
- Count Vector: occurrence of a word in a single document
- TF-IDF Vector: takes into account the entire corpus
- Co-Occurrence Vector: Similar words tend to occur together and will have similar context for example – Apple is a fruit. Mango is a fruit.

Prediction based learning-
- Language Model
- Word2Vec: A toolkit that allows the seamless training and use of pre-trained embeddings
  - CBOW
  - Skip-gram

Language Modelling: Bengio's Work in 2003

consists of a one-hidden-layer feed-forward neural network that predicts the next word in a sequence

Word2Vec: In 2013, Mikolov et al.proposed two architectures for learning word embeddings that are computationally less expensive than previous models
- CBOW: Continuous Bag of Words uses surrounding words to predict the centre word
- Skip Gram: uses the center word to predict the surrounding words

Function using Slack RTM API: Parses a list of events coming from the Slack RTM API to find bot commands. If a bot command is found, this function returns a tuple of command and channel.
Documentation for using bots on slack: Link

Logging to catch repeated offenders: The bot will log details of the repeat offenders and will send a message to the workspace admin. Regular offenders will also be shown and incremental 'testosterone level' value

Moving it into more workplace conversation apps like Microsoft Teams, Google Hangouts,

Audio Analysis: If the data gathered is significant enough, we can move this to audio/speech recognition of sexist comments. For example, this bot could be integrated with existing virtual assistants to give out warnings if a sexist remark is heard in a meeting.

The objective of the bot: Raise awareness of the toxic remarks in day-to-day forums used by people
Easily available toxic comment data
Used the same preprocessing steps
Used a classification model instead of LSTM
Future Work: Bots similar to Reddit bots: Analyzing the toxic nature of other social media platforms: Twitter, Forum channels

Beginner: Use the code in github for toxic comment and build it for reddit channels
Create Analysis of your reports: What should people learn from it?

Please feel free to push any sort of data or remark you find suitable for this project. I believe that sexism in workplace conversations is a global problem. I hope that this bot will be a step towards eliminating it. Therefore, I need crowdsourced to improve the model and make it more usable

Crowdsourcing data
Unfortunately, there are no proper datasets for casual sexist remarks at workplaces. The available datasets that I have come across are extremely vulgar/obscene. However, sexist remarks in workplaces are often subtle or contain some usual phrases like "Girls are like that".

Feedback- A link to give me a feedback- https://tinyurl.com/bots-to-combat-sexism
I would love to hear from you and your responses will be Anonymous!
You can get a link to the slides right after filling the feedback form!
Contact us:
Archana Iyer:
- varchanaiyer139@gmail.com