Spaces:
Running
Running
| import streamlit as st | |
| import torch | |
| import pickle | |
| import time | |
| import pandas as pd | |
| from iteration_utilities import unique_everseen | |
| from sentence_transformers import util | |
| from loader import bi_encoder, cross_encoder, df, job_corpus_ecoded, job_corpus | |
| def jobsearch(query,df, top_k=100): | |
| #print("Answer by NinjaBot : ") | |
| ans = [] | |
| question_embedding = bi_encoder.encode(query, convert_to_tensor=True) | |
| hits = util.semantic_search(question_embedding, job_corpus_ecoded, top_k=top_k) | |
| hits = hits[0] | |
| cross_inp = [[query, job_corpus[hit['corpus_id']]] for hit in hits] | |
| cross_scores = cross_encoder.predict(cross_inp) | |
| for idx in range(len(cross_scores)): | |
| hits[idx]['cross-score'] = cross_scores[idx] | |
| hits = sorted(hits, key=lambda x: x['cross-score'], reverse=True) | |
| #indexes = [] | |
| search_result = [] | |
| for idx, hit in enumerate(hits[0:10]): | |
| obj = {} | |
| ans.append(job_corpus[hit['corpus_id']]) | |
| #indexes.append(job_corpus.index(job_corpus[hit['corpus_id']])) | |
| obj['title'] = df.at[job_corpus.index(job_corpus[hit['corpus_id']]),'title'] | |
| obj['link'] = df.at[job_corpus.index(job_corpus[hit['corpus_id']]),'url'] | |
| search_result.append(obj) | |
| final_search_result = list(unique_everseen(search_result)) | |
| return final_search_result | |
| #return df.at[indexes[0],'title'],df.at[indexes[1],'title'],df.at[indexes[2],'title'],df.at[indexes[3],'title'],df.at[indexes[4],'title'] | |
| #return ans[0],ans[1],ans[2],ans[3],ans[4] | |
| def main(): | |
| if 'submitted' not in st.session_state: | |
| st.session_state.submitted = False | |
| def callback(): | |
| st.session_state.submitted = True | |
| st.title('Job Search Engine 💼') | |
| st.text("") | |
| st.text("") | |
| query = st.text_input('Enter your job query here ! ') | |
| if (st.button("Search", on_click=callback) and query) : | |
| with st.spinner('Fetching the best jobs for you!...'): | |
| #time.sleep(10) | |
| result = jobsearch(query, df) | |
| #result = jobsearch(query, df) | |
| st.success('NinjaBot : Here are a few suggestions') | |
| #st.write(f"This is the query : {query}") | |
| st.write(result) | |
| main() | |