app.py

from sentence_transformers import SentenceTransformer
import numpy as np
import requests

import gradio as gr
from description import project_description

def request_and_response(url):
    response = requests.get(url)
    papers = response.json()
    return papers

def extract_abstracts_and_ids(papers):
    abstracts = [paper["paper"]["summary"] for paper in papers]
    paper_ids = [paper["paper"]["id"] for paper in papers]
    return abstracts, paper_ids

def get_embeddings(model, texts):
    embeddings = model.encode(texts)
    return embeddings

def compute_similarity(model, embeddings1, embeddings2):
    similarities =  model.similarity(embeddings1, embeddings2)
    return similarities

def find_closest(similarities, paper_ids):
    best_match_idx = np.argmax(similarities)
    best_match_id = paper_ids[best_match_idx]
    return best_match_id

# Step 0: Get the model
model = SentenceTransformer("all-MiniLM-L6-v2")

# Step 1: Get papers from API
papers = request_and_response("https://hf.co/api/daily_papers")

# Step 2: Extract abstracts and paper ids
abstracts, paper_ids = extract_abstracts_and_ids(papers)

# Step 3: Embed Query and the Abstracts of papers
abstract_embeddings = get_embeddings(model, abstracts)

def get_closest_paper(query):
    query_embeddings = get_embeddings(model, [query])

    # Step 4: Find similarity scores
    similarities = compute_similarity(model, query_embeddings, abstract_embeddings)

    # Step 5: Find the closest match
    best_match_id = find_closest(similarities, paper_ids)

    # Step 6: Get the best match paper title and id
    paper = request_and_response(f"https://hf.co/api/papers/{best_match_id}")
    title = paper["title"]
    summary = paper["summary"]

    return title, f"https://huggingface.co/papers/{best_match_id}", summary


with gr.Blocks() as iface:
    gr.Markdown(project_description)
    with gr.Row():
        with gr.Column(scale=1):
            query = gr.Textbox(placeholder="What do you have in mind?")
            btn = gr.Button(value="Submit")
        with gr.Column(scale=3):
            with gr.Row():
                title = gr.Textbox()
                paper_link = gr.Textbox()
                abstract = gr.Textbox()
    
    btn.click(get_closest_paper, query, [title, paper_link, abstract])



if __name__ == "__main__":
    iface.launch()