Culture FitT -- The Idea
Culture FitT_(witter)_ is a tweet analyzing tool to help recruiters see if a prospective employee fits into the culture of the company.
Our team wanted to do something different than the stereotypical social media analyzing tool -- which looks for negative aspects of an online profile and "red-flags." We decided to gear toward the positive things that comes out of a person's profile. We chose Twitter to analyze this information because we believe want a person chooses to say when they're under a constraint holds more weight and is more revealing of a person's true character.
As Mark Twain once put it _ “I didn't have time to write a short letter, so I wrote a long one instead.” _
Back End We used Node.js and express in the back end (Mark and Kerlin) for the server and as a means of communication between R files written by Krista and the server. Various node modules were implemented alongside our R code in order to compliment each other. We used the twitter API in order to get all of a user's tweets and info.
Data Data Data We did the majority of the data analysis in R we had to use natural language processing and parse through lots of tweets to get the information we needed to extract relevant data. We used this data to build WorldCloud graphs, determine the most important interactions between other users, etc. We supplemented the R data analysis with some classifications done using Node.js to determine attributes of the profile, pull sentiment data, analysis of passive vs. active voice, and determine professionalism.
Front End and Design
We used semantic UI for the front end frame work, jQuery for front end manipulation, moment.js for time manipulation. Google Fonts for pretty text.
For the Designs we used PhotoShop and Illustrator.
How does it work
The user Enters a twitter screen name they want to have analyzed into our web-application. This name is sent to our server so we can pull the information from their twitter. This information can be pulled as long as you are following them on twitter or they are public. Our server pulls the information, analyzes the data, and returns valuable insights and graphics to the user.
Mark This is the first time Mark ever did back end. Normally in hackathons he handles front end work. He learned how to use API's, how to use OAuth, and how to communicate with other languages(specifically R) using child processes in Node.js.
Roshi Got to work with new features in illustrator called the "shaper tool."
Krista Krista learned a lot about natural language processing. She learned how to parse through large amounts of data and format it an easily accessible way. She also learned what Perl was, and was able to use it to assist in parsing text and the natural language processing. This is what allowed us to make the WordClouds. She gained deeper understanding of making Graphs in R, and how to write them to a file in the command line. Krista also gained more experience with R packages like, tm, worldcloud, jsonlite, and ggplot2. This was the first hackathon that she was successfully able to interpret json files into R.
Kerlin Kerlin also learned natural language processing. Got experience running R in Node.js and exposure to R. He also learned beginning data analysis and data manipulation. He also learned how to use a basic machine learning technique called classification.