def find_similar_pairs(data):
"""
Find the most similar document for each document in the collection, output the pairs
"""
processor = TextProcessor()
processor.map_json_data(data)
similar_pairs = processor.similarity_analysis()
data_output = {
processor.doc_collection[f].link: processor.doc_collection[s].link
for f, s in similar_pairs
}
with open('similar_0817.json', 'w') as file_output:
json.dump(data_output, file_output)
"DATA_PATH": str,
"session_num": 13,
}
if os.environ["HOME"] == "/root":
args["DATA_PATH"] = "/content/gdrive/MyDrive/bert-for-hmltc/data"
else:
args["DATA_PATH"] = "data"
random.seed(args["seed"])
np.random.seed(args["seed"])
torch.manual_seed(args["seed"])
logger.info("Initializing…")
tokenizer = load_tokenizer(args)
processor = TextProcessor(args, tokenizer, logger, "topic_list.json")
if args["use_parents"]:
model = create_experimental(args, len(processor.labels))
else:
model = create_baseline(args, len(processor.labels))
model_state_dict = torch.load(
join(args["DATA_PATH"], "model_files/13_finetuned_pytorch_model.bin"),
map_location="cpu",
)
model.load_state_dict(model_state_dict)
if args["do_train"]:
trainer = ModelTrainer(args, model, logger)
logger.info("Loading data…")
from flask import Flask, render_template, request
from processor import TextProcessor, Document
import re
from bs4 import BeautifulSoup
import urllib3
import numpy as np
from flask.ext.pymongo import PyMongo
import os
import sys
app = Flask(__name__)
app.config['MONGO_URI'] = os.getenv('MONGOHQ_URL')
mongo = PyMongo(app)
processor = TextProcessor()
with app.app_context():
processor.map_data(mongo.db.postings.find())
processor.build_doc_matrix()
@app.route('/', methods=['GET', 'POST'])
def main():
if request.method == 'GET':
return render_template('index.html')
url = request.form['url'].strip()
domain = 'newyork.craigslist.org/'
if not domain in url:
return render_template('index.html', error='Please enter a valid URL')
def setUp(self):
self.processor = TextProcessor()
