Technical Glossary
This glossary explains the technical terms that appear throughout the SisterShield documentation in plain language. You don’t need a computer science background to follow along — each term is described in everyday words.
Frameworks & Tools
Next.js
The main framework used to build SisterShield. It handles both what you see in the browser and the behind-the-scenes server work, all in one project.
App Router
The way Next.js organizes pages. Each folder in the app/ directory becomes a URL path (like /dashboard or /courses), making it easy to see how the site is structured just by looking at the folders.
Server Components
Parts of the page that are built on the server before being sent to your browser. This makes pages load faster because less code needs to run on your device.
API Routes
Special URLs on the server that handle behind-the-scenes work like saving data, logging in, or generating AI content. The browser sends a request, and the API route sends back a result (usually as JSON data).
Middleware
Code that runs automatically before every page loads. SisterShield uses it to detect which language you prefer and to check if you’re logged in before showing protected pages.
Prisma
A tool that lets you talk to the database using regular code instead of writing raw database commands. It also keeps the database structure in sync with the app, so when you add a new field in code, the database updates too.
ORM
ORM (Object-Relational Mapping) is a way to use your database through normal code instead of special database language (SQL). Think of it as a translator between your app and the database.
PostgreSQL
A free, reliable database that stores all the app’s data — users, courses, progress, documents. Think of it like a giant organized spreadsheet that lives on the server and can handle many users at once.
NextAuth
A ready-made login system for Next.js apps. It handles signing in, keeping you logged in across pages, and checking who you are so the app can show you the right content for your role.
JWT
JWT (JSON Web Token) is a small digital pass stored in your browser (as a cookie) that proves you’re logged in. The server checks this pass on every request so you don’t have to type your password again each time.
shadcn/ui
A set of pre-built interface pieces — buttons, dialogs, dropdown menus, tabs, progress bars, etc. They look good, work on phones and desktops, and are accessible to people using screen readers.
Tailwind CSS
A styling system where you design pages by adding short class names directly in your HTML. For example, text-lg makes text bigger, p-4 adds spacing, and flex arranges items in a row. No separate CSS files needed.
next-intl
The library that makes SisterShield work in both English and Korean. It loads the right translations for the user’s chosen language and handles switching between languages smoothly.
Twine & Interactive Fiction
Twine
A free tool for building choose-your-own-adventure stories. You write scenes (called “passages”) and connect them with clickable links so readers can make choices that change the story.
TwinePlayer
The part of SisterShield that plays Twine stories inside the app. It loads the story in a secure frame and tracks which passages the student has visited and how far along they are.
Twee
A simple text format for writing Twine stories by hand. You type :: to start a new scene and [[brackets]] to create clickable choices. The AI writes stories in this format, and then they get compiled into playable HTML.
Tweego
A tool that turns Twee text files into playable Twine stories you can open in a browser. Think of it as a compiler that converts the plain-text script into a finished interactive experience.
Harlowe
The default story engine for Twine 2. It controls how stories look and behave, including things like tracking variables (e.g., a score), if/else branching, and text styling.
SugarCube
An alternative Twine story engine with more advanced features like save/load, custom JavaScript code, and complex game-like mechanics. It’s more powerful but also more complex.
postMessage
A way for the Twine story (running inside its own frame) to send messages back to the main SisterShield page — for example, “the student just finished this passage” or “the quiz score is 3/3.”
MutationObserver
A browser feature that watches for changes on a page. SisterShield uses it to detect when a student clicks to a new passage in a Twine story, so it can update the progress tracker.
AI & Machine Learning
LLM
LLM (Large Language Model) is an AI that can read and write human-like text — like Claude or ChatGPT. SisterShield uses LLMs to generate interactive stories, translate content between languages, and fix story errors.
Claude
An AI made by Anthropic. SisterShield uses Claude to write educational stories, translate text between English and Korean, and automatically fix story structure problems.
GPT-4o
An AI made by OpenAI that can understand both text and images. SisterShield can use it as an alternative to Claude for writing stories and translations.
DALL-E 3
An AI by OpenAI that creates pictures from written descriptions. SisterShield uses it to make cover art and scene illustrations for courses, with built-in safety rules to keep images appropriate.
Embedding
A way to turn words into a list of numbers that capture their meaning. Sentences about similar topics get similar numbers, which is how the AI search finds documents related to what you’re looking for — even if the exact words are different.
text-embedding-3-small
A specific OpenAI model that turns text into a list of 1,536 numbers representing its meaning. SisterShield uses it to convert documents and search queries into these number lists so it can find the most relevant content.
Cosine Similarity
A math formula that measures how similar two pieces of text are by comparing their number lists (embeddings). A score of 1 means basically identical, 0 means completely unrelated. It’s how the AI decides which documents are most relevant to your question.
RAG Pipeline
RAG
RAG (Retrieval-Augmented Generation) is a technique where the AI first looks up real facts from a knowledge base, then uses those facts when writing. Instead of making things up, it bases its answers on actual source documents — like a student who checks their textbook before answering a question.
Hybrid Search
Searching two ways at once: by meaning (“do these topics match?”) and by exact keywords (“does this specific word appear?”). Combining both gives better results than either method alone.
RRF
RRF (Reciprocal Rank Fusion) is a way to merge results from two different search methods into one best-of-both list. If a document ranks high in both meaning search and keyword search, it gets pushed to the top.
pgvector
An add-on for the PostgreSQL database that lets it search by meaning, not just exact words. It stores the numerical representations (embeddings) of documents and can quickly find the most similar ones.
IVFFlat
A speed trick for searching large amounts of data. Instead of comparing your query against every single document (which is slow), it groups similar items into clusters and only checks the nearest groups — much faster with only a tiny accuracy trade-off.
RagDocument
A database record that keeps track of an uploaded file — its name, what topic category it belongs to (like TF-CBT or TF-VAWG), and whether it’s been processed yet.
RagChunk
A small piece of a larger document, stored in the database along with its numerical meaning (embedding). When the AI needs facts, it searches these chunks to find the most relevant paragraphs.
RagConceptTag
A label that connects a document chunk to a topic (like “cyberbullying” or “coping skills”). These labels make it easy to find all chunks about a specific subject.
RagCitation
A record that tracks exactly which source documents the AI used when writing a story. This lets you click through to see where the information originally came from, keeping the AI transparent.
KnowledgeConcept
A topic in the system’s organized knowledge tree — like “online harassment,” “legal protections,” or “coping skills.” Topics are arranged in a parent-child hierarchy so related concepts are grouped together.
Concept Tagger
An automatic process that reads each document chunk and labels it with the topics it covers. Think of it like auto-tagging photos with keywords — but for educational documents.
File Formats
The common document format you see when downloading reports, research papers, or official documents. SisterShield can read PDFs and add their content to the AI knowledge base.
DOCX
Microsoft Word’s file format (the files ending in .docx). SisterShield can read Word documents and add their content to the knowledge base, just like PDFs.
SHA-256
A math function that creates a unique fingerprint for any file. If even one character in the file changes, the fingerprint changes completely. SisterShield uses this to know when a document has been updated and needs to be re-processed.
Domain-Specific
TF-VAWG
TF-VAWG (Technology-Facilitated Violence Against Women and Girls) means harmful behaviors carried out through technology — like cyberbullying, online stalking, sharing someone’s private images without consent, or using apps to control or monitor someone.
TF-CBT
TF-CBT (Trauma-Focused Cognitive Behavioral Therapy) is a type of therapy designed to help young people recover from traumatic experiences. It uses structured, evidence-based techniques developed by mental health professionals.
i18n
i18n (short for “internationalization” — there are 18 letters between the “i” and the “n”) means making software work in multiple languages. SisterShield supports English and Korean, with every button, label, and message available in both.
bcrypt
A tool that scrambles passwords into unreadable text before storing them. Even if someone accessed the database, they couldn’t read the original passwords. SisterShield uses this to keep user accounts safe.
sessionStorage
A temporary storage area in your browser that holds small bits of data (like which tab you had open) while you’re on the site. SisterShield clears this when you use the Quick Exit button so no trace of your activity remains.
GDPR
GDPR (General Data Protection Regulation) is a European law that gives people rights over their personal data — like the right to download it or delete their account. SisterShield is planning to add these features.