Fellow

EECS 371: Knowledge Representation and Reasoning | W2019

Andres Kim, Veronica Medrano, Kapil Garg

Fellow creates a querying and reasoning structure that allows students to find other like-minded individuals based on academic year, program, career goals, hobbies, and software experience. The long-term goal for this project is to build a web app that can be interacted with not only to find similar students, but also professor, university organizations, events, and more.

Note: the current repository already contains the final generated KRF file in dist/krf/fellow.krf. To load it and run it in Companion, feel free to skip to the section Running Companion.

Directory Structure

├── dev (development code)
│   ├── krf (intermediate krf files)
│   │   ├── facts.krf (defines entities, collections, and relations)
│   │   ├── people.krf (defines people entities scraped from survey)
│   │   ├── queries.krf (defines horn clauses for queries)
│   │   ├── query_definitions.krf (defines queries)
│   │   └── test-queries.krf (some included test queries to play with)
│   └── scraping
│       ├── Person.py (model for a Person)
│       ├── generate_krf.py (generates the combined dist/krf/fellow.krf file using the intermediate krf files)
│       └── sheet_scraping.py (generates the dev/people.krf by scraping the Google Sheet survey responses)
└── dist (distribution code)
    └── krf
        └── fellow.krf (combined krf file to use with Companions)

Setup

Pipenv

We use pipenv for managing package dependencies.

1. Install pipenv using the link above.
2. Run pipenv install to install package dependencies and pipenv shell to start virtual environment with installed dependencies.

Credentials

You will need Google authentication credentials in order to access Google Sheets.

1. Follow the instructions from the gspread repository to get a credential.json file.
2. Put your credential.json file in dev/scraping.

.env File

We pull the spreadsheet URL from a .env file that is loaded into the pipenv virtual environment when you run pipenv shell. Generate one of these files using the format below and place it in the root of the cloned repository:

SHEET_URL=YOUR_GOOGLE_SPREADSHEET_URL

Generating KRF File

1. Navigate to dev/scraping/
2. Run python generate_krf.py to scrape data from the Google Sheet and generate a single krf file for Companion.
3. Output file can be found in dist/krf.

Running Companion

1. Startup companion and load the dist/krf/fellow.krf flat file into the interaction manager.
2. Browse the agent of the interaction manager to input queries and explore the knowledge base.
3. To make sure everything is loaded correctly, try running the following query:

(sameAcademicYear DestinyLivingston ?suggestedPerson)

This should output students that have the same academic year as Destiny Livingston. Note: you must run the queries using the KRR-Winter2019FactsMt microtheory.

Available Queries

Feel free to test more of the available queries and compose custom ones.

Who is in the same academic program as me?

(sameAcademicProgram ?person ?suggestedPerson)

Example Query: (sameAcademicProgram DestinyLivingston ?suggestedPerson)

Who is in the same academic year as me?

(sameAcademicYear ?person ?suggestedPerson)

Example Query: (sameAcademicYear DestinyLivingston ?suggestedPerson)

Who is in the same field of work as me?

(sameFieldOfWork ?person ?suggestedPerson)

Example Query: (sameFieldOfWork DestinyLivingston ?suggestedPerson)

Who has the same software experience as me?

(sameSoftwareExperience ?person ?suggestedPerson)

Example Query: (sameSoftwareExperience DestinyLivingston ?suggestedPerson)

Who has the same academic interests as me?

(sameAcademicInterests ?person ?suggestedPerson)

Example Query: (sameAcademicInterests DestinyLivingston ?suggestedPerson)

Who has the same hobbies as me?

(sameHobbies ?person ?suggestedPerson)

Example Query: (sameHobbies DestinyLivingston ?suggestedPerson)

What are some interesting events that would be related to me?

(interestingEvents ?person ?event)

Example Query: (interestingEvents DestinyLivingston ?event)

Who is someone who is experienced in a specific software?

(personExperiencedWith ?suggestedPerson ?program)

Example Query: (personExperiencedWith ?suggestedPerson GameDevelopment)

Who is similar to me?

(similarNUPerson ?person ?suggestedPerson)

Example Query: (similarNUPerson DestinyLivingston ?suggestedPerson)

Pre-defined Composite Queries

The queries below are compositions of the basic queries listed above:

(sameHobbiesAndYear ?person ?suggestedPerson)

(sameMajorAndYear ?person ?suggestedPerson)

(sameHobbiesAndField ?person ?suggestedPerson)

(sameHobbiesWithExperienceIn ?person ?suggestedPerson ?program)

Custom Composite Queries

We define 3 custom queries that take the simple queries from earlier as input, along with a person to match to. These are custom queries that can be used to run any combination of simpler queries.

Example Query: (customQuery2 sameAcademicInterests sameHobbies DestinyLivingston ?suggestedPerson)

(customQuery2 ?query1 ?query2 ?person ?suggestedPerson)

(customQuery3 ?query1 ?query2 ?query3 ?person ?suggestedPerson)

(customQuery4 ?query1 ?query2 ?query3 ?query4 ?person ?suggestedPerson)

Customizing and Adding More Data

1. If you want to create a new entity representing a NU student in the knowledge base, please check dev/krf/people.krf
2. Follow the structure and syntax, replacing information as needed.
3. To facilitate the process, append the newly created data to the bottom of dist/krf/fellow.krf and reload it in Companion.
4. Now try running some queries related to your new user like such:

(sameAcademicYear newUser ?suggestedPerson)