app.py

import streamlit as st
from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity
import openai
import numpy as np
import pandas as pd

# App title
st.title("Keyword Cosine Similarity Tool")

# Overview
st.markdown(
    """
    **Purpose:** 
    Quickly remove irrelevant keywords from your keyword research and move to the next step in your optimization!

    Have you ever had to review a long list of queries to determine whether they were relevant to your target keyword? This Space aims to automate that process by entering your primary keyword and a list of related queries from any source you might do keyword research.
    The resulting table is an ordered list of your comparison keywords based on the cosine similarity of each query's embeddings.

    **Instructions:**
    1. Enter your **Primary Keyword** in the input field.
    2. Provide a list of **Keywords to Compare** (separated by new lines or commas).
    3. Select an **Embedding Model** to compute keyword embeddings.
    4. If using OpenAI embeddings, input your **API Key**.
    5. Click **Calculate Similarities** to compute and rank your keywords by relevance.

    **Output:**
    - A sorted table of your comparison keywords based on their cosine similarity to your primary keyword.
    - Option to download the results as a CSV file.
    """
)

# Inputs
st.header("Input Parameters")
primary_keyword = st.text_input("Primary Keyword", placeholder="Enter your primary keyword")
st.text_area(
    "Keywords to Compare",
    placeholder="Enter keywords separated by new lines or commas",
    help="You can input keywords on separate lines or separated by commas. Any keywords with commas will be treated as separate queries.",
    key="keywords"
)
keywords = st.session_state.keywords
model_name = st.selectbox("Select Embedding Model", ["sentence-transformers/LaBSE", "sentence-transformers/all-MiniLM-L6-v2", "OpenAI Embeddings"])
openai_api_key = st.text_input("OpenAI API Key (optional)", type="password")

# Process Button
if st.button("Calculate Similarities"):
    if not primary_keyword or not keywords:
        st.error("Please provide both the primary keyword and keywords to compare.")
    else:
        # Process keywords: Split by commas and/or new lines
        keyword_list = [kw.strip() for kw in keywords.replace(",", "\n").split("\n") if kw.strip()]
        
        # Check model type
        if model_name in ["sentence-transformers/LaBSE", "sentence-transformers/all-MiniLM-L6-v2"]:
            # Load SentenceTransformer model
            st.info(f"Loading model: {model_name}")
            model = SentenceTransformer(model_name)
            
            # Generate embeddings
            st.info("Generating embeddings...")
            primary_embedding = model.encode(primary_keyword, convert_to_tensor=True)
            keyword_embeddings = model.encode(keyword_list, convert_to_tensor=True)
        
        elif model_name == "OpenAI Embeddings":
            if not openai_api_key:
                st.error("Please provide your OpenAI API key for this model.")
            else:
                openai.api_key = openai_api_key
                st.info("Generating OpenAI embeddings...")
                
                def get_openai_embedding(text):
                    response = openai.Embedding.create(
                        model="text-embedding-ada-002",
                        input=text
                    )
                    return np.array(response['data'][0]['embedding'])
                
                primary_embedding = get_openai_embedding(primary_keyword)
                keyword_embeddings = np.array([get_openai_embedding(kw) for kw in keyword_list])
        
        else:
            st.error("Invalid model selection.")
            st.stop()
        
        # Calculate cosine similarities
        st.info("Calculating cosine similarities...")
        similarities = cosine_similarity([primary_embedding], keyword_embeddings)[0]
        
        # Sort results by cosine similarity
        st.info("Sorting results...")
        results = [{"Keyword": kw, "Cosine Similarity": sim} for kw, sim in zip(keyword_list, similarities)]
        sorted_results = sorted(results, key=lambda x: x["Cosine Similarity"], reverse=True)

        # Display results
        st.header("Results")
        df_results = pd.DataFrame(sorted_results)
        st.table(df_results)

        # Download results as CSV
        st.download_button(
            label="Download Results as CSV",
            data=df_results.to_csv(index=False),
            file_name="cosine_similarity_results.csv",
            mime="text/csv"
        )

        # Debugging/Intermediate Data
        st.header("Debugging Info")
        st.write("Primary Embedding:", primary_embedding)
        st.write("Keyword Embeddings:", keyword_embeddings)

# Footer
st.markdown("---")
st.markdown("Created by [Ryland Bacorn](https://huggingface.co/ReithBjarkan). Report a [bug or make a suggestion](mailto:rybacorn@gmail.com)")