📘 IIS Project – Rails 6.1 + PostgreSQL + Vite + Bootstrap

This project is a miniature information-retrieval system featuring:

• Rails 6.1

• PostgreSQL

• Vite Ruby (JS bundler)

• Bootstrap 5

• Chart.js

• Ability to:

  • Import .trec files
  • Auto-compute style vectors (TTR, sentence length, pronoun ratio, readability)
  • Query documents via different scoring approaches
  • Visualize vectors in the browser

---

🚀 1. Requirements

Ensure the following are installed:

• Ruby 3.3.x
• Rails 6.1
• PostgreSQL 13+
• Node.js / Bun
• Yarn or npm
• Vite Ruby (installed via Gemfile)

---

🏗 2. Installation

Clone the project:

git clone https://github.com/your-org/iis_project.git
cd iis_project

Install Ruby dependencies:

bundle install

Install JavaScript dependencies (using Bun or Yarn):

bun install
# or:
yarn install

---

🛢 3. Database Setup

rails db:create
rails db:migrate

---

🎨 4. Frontend Setup – Vite Ruby

Start the Vite dev server:

bin/vite dev

If needed, restart the Rails server afterward.

The frontend entrypoint is:

app/frontend/entrypoints/application.js

---

📥 5. Importing a .trec File

Import any TREC XML file containing:

<DOC>
  <recordId>DOC001</recordId>
  <text>Full document text…</text>
  <style_vec>1.0,4.3,0.05,60.2</style_vec> <!-- optional -->
</DOC>

Run:

rake trec:import FILE=path/to/file.trec

The task will:

1. Parse the XML
2. Compute style vectors if missing (StyleFeatureService)
3. Store everything in PostgreSQL
4. Reindex Document model (Searchkick-style or custom)

---

📊 6. Style Vector Extraction

Style vectors use:

1. TTR (type-token ratio)
2. Avg sentence length
3. Pronoun ratio
4. Flesch readability score

Implemented in:

app/services/style_feature_service.rb

---

🔎 7. Search Modes

1. Lexical Search

GET /search?q=keyword

2. Style-based Similarity Search

GET /search_style?q=your text

3. Hybrid Search (lexical + vector score)

GET /search_hybrid?q=your text

---

🌐 8. Application Routes

Path	Purpose
/home	Document list
/search	Lexical search
/search_style	Vector-based search
/search_hybrid	Combined score
/vectors	Vector visualization UI

Route definitions:

get "home" => "documents#index", as: :documents_home
get "vectors" => "vectors#index"

get "search" => "documents#search"
get "search_style" => "documents#search_style"
get "search_hybrid" => "documents#search_hybrid"

---

📊 9. Vector Visualization UI

Accessible at:

/vectors

Displays:

• Document ID
• Title
• Body (optional)
• Extracted style vector
• Mini-charts using Chart.js inside Bootstrap Accordions

---

▶️ 10. Start the Application

In one terminal:

bin/vite dev

In another:

rails s

Then visit:

http://localhost:3000/home

---

📚 11. Directory Overview

app/models/document.rb             # Model storing trec_id, title, body, style_vec
app/services/style_feature_service.rb
app/controllers/documents_controller.rb
app/controllers/vectors_controller.rb
lib/tasks/trec_import.rake
app/frontend/entrypoints/application.js
app/views/vectors/index.html.haml  # Visualization UI

---

🏁 12. Summary

This project provides:

✔ Import + processing of TREC documents ✔ Style-feature vector extraction ✔ Multiple search strategies ✔ Beautiful Vite-powered frontend ✔ Chart.js vector visualization ✔ Clean Rails 6.1 backend