Add app files

README.md

@@ -1,13 +1,16 @@
 ---
 title: Legislation Search
-emoji: 💻
-colorFrom: gray
-colorTo: purple
+emoji: 🇧🇪
+colorFrom: yellow
+colorTo: yellow
 sdk: gradio
-sdk_version: 3.0.24
 app_file: app.py
 pinned: false
 license: cc-by-nc-sa-4.0
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+This is a demo for the paper: "A Statutory Article Retrieval Dataset in French" (ACL 2022)
+
+- Paper: https://aclanthology.org/2022.acl-long.468/
+- Code: https://github.com/maastrichtlawtech/bsard
+- Data: https://huggingface.co/datasets/antoiloui/bsard

app.py

@@ -0,0 +1,26 @@
+import gradio as gr
+import pandas as pd
+
+import torch
+from datasets import load_dataset
+from sentence_transformers import SentenceTransformer, util
+
+
+q_encoder = SentenceTransformer("checkpoints/q_encoder")
+doc_embeddings = torch.load('checkpoints/doc_embeddings.pt')
+docs = pd.DataFrame(load_dataset("antoiloui/bsard", data_files="articles_fr.csv")['train'])
+
+def search(query):
+    q_emb = q_encoder.encode(query, convert_to_tensor=True)
+    hits = util.semantic_search(q_emb, doc_embeddings, top_k=100, score_function=util.cos_sim)[0]
+    return {docs.loc[h['corpus_id'], 'article']: f"Art. {docs.loc[h['corpus_id'], 'article_no']}, {docs.loc[h['corpus_id'], 'code']}" for h in hits[:5]}
+
+gr.Interface(
+    fn=search,
+    inputs=gr.Textbox(label="Question", placeholder=""),
+    outputs=[gr.Textbox(lines=5, label="Result"),gr.Textbox(label="Reference")],
+    title="Legislation Search 🇧🇪",
+    description="",
+    flagging_options=["👍","👎"],
+    examples=["Qu'est-ce que je risque si je viole le secret professionnel ?", "Mon employeur peut-il me licencier alors que je suis malade ?"]
+).launch(share=False, enable_queue=False)

checkpoints/doc_embeddings.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dc0b8cb264f836bfa266ea91736b71acbdc16d4306708aa958392d283b55004
+size 69529323

requirements.txt

@@ -0,0 +1,4 @@
+pandas==1.3.5
+sentence-transformers==2.1.0
+torch==1.10.1
+datasets==1.18.3