Configure GitHub Linguist language statistics

Add .gitattributes rules to exclude non-core code from GitHub's language statistics. This includes archived dashboard backup code, IDE tooling, Jupyter notebooks, and build artifacts. Prevents these from misrepresenting the repository's primary languages (Python and JavaScript for the active application).

.gitattributes

@@ -1,4 +1,29 @@
+# .gitattributes - GitHub Linguist configuration for Multilingual-Absa
+
 # Auto detect text files and perform LF normalization
 * text=auto
+# ==========================================
+# EXCLUSIONS FROM LANGUAGE STATISTICS
+# 1. Archived/Backup Dashboard Code
+# The dashboard_backup directory contains a legacy Vite/React application
+# including large minified JS assets in dist/, CSS, and vendored node_modules.
+# We exclude this entire directory to prevent archived code from dominating statistics.
+dashboard_backup/** linguist-generated=true
+# 2. IDE and Tooling Hooks
+# The .opencode directory contains generated JS hooks for the workspace environment.
+# These are internal tooling scripts and not part of the active application source code.
+.opencode/** linguist-generated=true
+# 3. Jupyter Notebooks
+# Notebooks are used for exploration and training on Colab, but the core active
+# application is the Python backend and FastAPI. Excluding these prevents notebooks
+# from misrepresenting the primary languages.
+ml/notebooks/*.ipynb linguist-generated=true
+# 4. Standard Build Artifacts & Vendored Dependencies
+# Ensures that any inadvertently committed build artifacts or vendor libraries
+# (like node_modules in subdirectories) do not skew language statistics.
+dist/** linguist-generated=true
+build/** linguist-generated=true
+node_modules/** linguist-generated=true
+**/node_modules/** linguist-generated=true
 *.ftz filter=lfs diff=lfs merge=lfs -text
 *.arrow filter=lfs diff=lfs merge=lfs -text