Decomp website

This repository contains the source for the Decompositional Semantics Initiative website. It uses the So Simple Theme by Michael Rose. To learn how to install and use this theme check out the Setup Guide.

Modifying site (meta)data

As a member of Decomp, there are two sorts of (meta)data you may want to add to the site: (meta)data about a researcher or (meta)data about a project.

Adding a researcher

There two steps necessary to add a researcher such that they auto-populate on pages for projects they have contributed to:

1. Create an entry for the researcher in the _data/researchers.yml file within the relevant group's people hash with a key built from the first letter of the researcher's first name plus their last name and a metadata hash with five entries: name, affiliation, website, email, and projects. For instance:

bvandurme:
  name: Benjamin Van Durme
  affiliation: Johns Hopkins University
  website: http://www.cs.jhu.edu/~vandurme/
  email: vandurme@cs.jhu.edu
  projects:
  - predpatt
  - spr
  - factuality
  - genericity
  - time
  - wsd

Note that all entries are string-valued, except projects, which maps to a list of strings. The strings contained within projects are handles associated with each project. It is important that these handles match the keys found in the top-level hash in _data/projects.yml, because the project list is used to populate the Researchers section of project pages. For instance, because spr is found in Ben's projects list and because the spr key in _data/projects.yml is associated with the Semantic Proto-Roles project, his avatar is part of the researchers section of the Semantic Proto-Roles project page.

2. Place a square .jpg image of the researcher into the images/ directory with a filename that matches that researcher's handle. For instance, images/bvandurme.jpg is the image associated with Benjamin Van Durme. Please make sure this .jpg uses the .jpg extension, not the .jpeg extension.

Adding a project

There are four steps to adding a project to the Projects and Data pages.

1. Create an entry for the project in the _data/projects.yml file with a metadata hash containing three entries: title, url, and either repository (for code projects) or datasets (for annotation projects). For instance, the PredPatt project is a code project, so it has title, url, and repository.

predpatt:
  title: PredPatt
  url: predpatt/
  repository: https://github.com/hltcoe/PredPatt

In contrast, the Semantic Proto-Roles project is an annotation project, so its hash contains the keys title, url, and datasets.

spr:
  title: Semantics Proto-Roles
  url: semantic-proto-roles/
  datasets:
    - name: v1
      filetype: tar.gz
      corpus: Penn TreeBank
      url: protoroles_eng_pb.tar.gz
      trainsize: XXXXX
      devsize: XXXXX
      testsize: XXXXX
    - name: v2
      filetype: tar.gz
      corpus: English Web TreeBank
      url: protoroles_eng_udewt.tar.gz
      trainsize: XXXXX
      devsize: XXXXX
      testsize: XXXXX

Note that all entries for an annotation project are string-valued, except datasets, which maps to a list of hashes. The hashes contained within projects contain metadata that are used to populate the table in the Data page. (A project is only added to the data page if it has a nonempty datasets list.) Each hash must have seven entries: name, filetype, corpus, url, trainsize, devsize, and testsize.

2. Create a directory for the project with a name that matches the value of url you entered into _data/projects.yml. For instance, the directory associated with the Semantic Proto-Roles project would be named semantic-proto-roles.

3. Create a file index.md within the directory you just made. This will be the project page. At the top of the file place the following metadata – where <HANDLE> is replaced with the project's handle from _data/projects.yml.

---
layout: project
handle: <HANDLE>
search_omit: true
---

For instance, the project page associated with the Semantic Proto-Roles project would have the following metadata.

---
layout: project
handle: spr
search_omit: true
---

After the metadata, include a description of the project along with citations. The data/code links and researcher section will be automatically populated from the project metadata in _data/projects.yml.

4. If the project is an annotation project, place any datasets associated with that project in the project directory—i.e. the same directory containing the project page.