mirror of
https://github.com/SkalaraAI/management-llm.git
synced 2025-04-03 20:10:20 -04:00
added model & matching algo to flask server
This commit is contained in:
Atin Pothiraj
parent
6a4c034f10
commit
403a489176
|
|
@ -3,7 +3,8 @@ import pandas as pd
|
|||
import matplotlib.pyplot as plt
|
||||
import os
|
||||
import openai
|
||||
|
||||
import joblib
|
||||
import random
|
||||
|
||||
app = Flask(__name__)
|
||||
openai.api_key = os.getenv("OPENAI_API_KEY")
|
||||
|
|
@ -12,67 +13,116 @@ openai.api_key = os.getenv("OPENAI_API_KEY")
|
|||
tasks_data = {}
|
||||
users_data = {}
|
||||
|
||||
def generate_prompt(task_form):
|
||||
base_text = "You are a software engineer at a software development company. Your job is to assign tags to tasks based on the software/tools used. Please generate one-word tags representing software/tools/libraries commonly used by developers to build or complete the following task."
|
||||
return base_text + "TASK: " + task_form
|
||||
# Loads the SVM model, vectorizer, and encoder
|
||||
model_filename = "svm_model.sav"
|
||||
svm_model = joblib.load(model_filename)
|
||||
|
||||
vectorizer_filename = 'fitted_vectorizer.joblib'
|
||||
word_vectorizer = joblib.load(vectorizer_filename)
|
||||
|
||||
encoder_filename = 'fitted_encoder.joblib'
|
||||
encoder = joblib.load(encoder_filename)
|
||||
|
||||
def generate_technical_tags(task_form):
|
||||
prompt = generate_prompt(task_form)
|
||||
|
||||
response = openai.Completion.create(
|
||||
model="text-davinci-003",
|
||||
prompt=prompt,
|
||||
temperature=0,
|
||||
max_tokens=100,
|
||||
stop=["TASK:"]
|
||||
)
|
||||
# Vectorize
|
||||
vectorized_text = word_vectorizer.transform([task_form])
|
||||
decision_function_scores = svm_model.decision_function(vectorized_text)
|
||||
|
||||
tags = response.choices[0].text.strip().split("\n")
|
||||
# Get the top 3 predicted labels based on highest decision function scores
|
||||
top_4_indices = decision_function_scores.argsort()[0][-3:][::-1]
|
||||
top_4_labels = encoder.inverse_transform(top_4_indices)
|
||||
|
||||
tags = top_4_labels.tolist()
|
||||
return tags
|
||||
|
||||
def match_tasks_to_users(tasks, users, user_max_tasks):
|
||||
task_ids = list(tasks.keys())
|
||||
|
||||
# Create a dictionary to store the matched tasks and users
|
||||
task_to_user_matches = {}
|
||||
user_to_task_matches = {user_id: None for user_id in users}
|
||||
|
||||
# Helper function to calculate the preference score between a task and a user
|
||||
def calculate_preference(task_id, user_id):
|
||||
task_tags = set(tasks[task_id]['tags'])
|
||||
user_strengths = set(users[user_id]['strengths'])
|
||||
user_current_tasks = users[user_id]['current_tasks']
|
||||
|
||||
# Calculate the preference score based on matching tags and strengths
|
||||
tag_score = len(task_tags.intersection(user_strengths))
|
||||
|
||||
# Calculate the total preference score, considering currentTasks as a tiebreaker
|
||||
preference_score = tag_score - user_current_tasks * 0.1
|
||||
|
||||
# Makes the preference score negative if the user is already at max tasks
|
||||
if user_current_tasks == user_max_tasks:
|
||||
preference_score -= user_max_tasks
|
||||
|
||||
return preference_score
|
||||
|
||||
# Assign tasks to users based on preferences
|
||||
for task_id in task_ids:
|
||||
task_info = tasks[task_id]
|
||||
|
||||
# Sort the users based on their preference for this task
|
||||
sorted_users = sorted(users.keys(), key=lambda user_id: calculate_preference(task_id, user_id), reverse=True)
|
||||
# Assign the task to the first user in the sorted list who is not already matched
|
||||
for user_id in sorted_users:
|
||||
if user_to_task_matches[user_id] is None:
|
||||
task_to_user_matches[task_id] = user_id
|
||||
user_to_task_matches[user_id] = task_id
|
||||
break
|
||||
print(task_to_user_matches)
|
||||
return task_to_user_matches
|
||||
|
||||
@app.route('/label_tasks', methods=['POST'])
|
||||
def get_task_tags():
|
||||
# try:
|
||||
data = request.get_json()
|
||||
tasks = data[0]['tasks']
|
||||
result = {}
|
||||
for task, task_description in tasks.items():
|
||||
task_id = task_description["id"]
|
||||
task_content = task_description["content"]
|
||||
task_complexity = task_description["complexityScore"]
|
||||
|
||||
tags = generate_technical_tags(task_content) # where the function that gets the tags is placed
|
||||
result[task_id] = tags
|
||||
return jsonify(result)
|
||||
# except Exception as e:
|
||||
# return jsonify({"error": str(e)}), 500
|
||||
|
||||
|
||||
@app.route('/assign_tasks', methods=['POST'])
|
||||
def assign_tasks_to_users():
|
||||
try:
|
||||
data = request.get_json()
|
||||
tasks = data[0]['tasks']
|
||||
users = data[0]['users']
|
||||
result = {}
|
||||
|
||||
tasks_dict = {}
|
||||
users_dict = {}
|
||||
|
||||
for task, task_description in tasks.items():
|
||||
tasks_dict[task_description["id"]] = {"tags" : task_description["content"], "complexity" : task_description["complexityScore"]}
|
||||
# tasks_dict[task_description["id"]] = {"tags" : generate_technical_tags(task_description["content"]), "complexity" : task_description["complexityScore"]}
|
||||
|
||||
for user, user_description in users.items():
|
||||
users_dict[user_description["id"]] = {"strengths" : user_description["strengths"], "current_tasks" : user_description["currentTasks"]}
|
||||
|
||||
|
||||
|
||||
# return jsonify(result)
|
||||
return result
|
||||
task_id = task_description["id"]
|
||||
task_content = task_description["content"]
|
||||
task_complexity = task_description["complexityScore"]
|
||||
|
||||
tags = generate_technical_tags(task_content) # where the function that gets the tags is placed
|
||||
result[task_id] = tags
|
||||
return jsonify(result)
|
||||
except Exception as e:
|
||||
return jsonify({"error": str(e)}), 500
|
||||
|
||||
|
||||
@app.route('/assign_tasks', methods=['POST'])
|
||||
def assign_tasks_to_users():
|
||||
# try:
|
||||
data = request.get_json()
|
||||
tasks = data[0]['tasks']
|
||||
users = data[0]['users']
|
||||
result = {}
|
||||
|
||||
tasks_dict = {}
|
||||
users_dict = {}
|
||||
|
||||
for task, task_description in tasks.items():
|
||||
tasks_dict[task_description["id"]] = {"tags" : generate_technical_tags(task_description["content"]), "complexity" : task_description["complexityScore"]}
|
||||
|
||||
for user, user_description in users.items():
|
||||
users_dict[user_description["id"]] = {"strengths" : user_description["strengths"], "current_tasks" : user_description["currentTasks"]}
|
||||
|
||||
|
||||
matching = match_tasks_to_users(tasks_dict, users_dict, 3)
|
||||
# result = {}
|
||||
# for task, task_description in tasks.items():
|
||||
# result[task_description["content"]] = {"assignedTo" : matchings[task_description["id"]]}
|
||||
|
||||
# print(result)
|
||||
result = matching
|
||||
return jsonify(result)
|
||||
# return result
|
||||
# except Exception as e:
|
||||
# return jsonify({"error": str(e)}), 500
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
app.run(debug=True)
|
||||
|
|
|
|||
BIN
task-server/fitted_encoder.joblib
Normal file
BIN
task-server/fitted_encoder.joblib
Normal file
Binary file not shown.
BIN
task-server/fitted_vectorizer.joblib
Normal file
BIN
task-server/fitted_vectorizer.joblib
Normal file
Binary file not shown.
28
task-server/model_integration_testing.py
Normal file
28
task-server/model_integration_testing.py
Normal file
|
|
@ -0,0 +1,28 @@
|
|||
import joblib
|
||||
|
||||
# To be translated
|
||||
sample_text = "Implement a login page with form validation using React.js."
|
||||
|
||||
# Loads the model, vectorizer, and encoder
|
||||
model_filename = "svm_model.sav"
|
||||
svm_model = joblib.load(model_filename)
|
||||
|
||||
vectorizer_filename = 'fitted_vectorizer.joblib'
|
||||
word_vectorizer = joblib.load(vectorizer_filename)
|
||||
|
||||
encoder_filename = 'fitted_encoder.joblib'
|
||||
encoder = joblib.load(encoder_filename)
|
||||
|
||||
# Vectorize
|
||||
vectorized_text = word_vectorizer.transform([sample_text])
|
||||
|
||||
binary_predictions = svm_model.predict(vectorized_text)
|
||||
|
||||
decision_function_scores = svm_model.decision_function(vectorized_text)
|
||||
|
||||
# Get the top 3 predicted labels based on highest decision function scores
|
||||
top_4_indices = decision_function_scores.argsort()[0][-3:][::-1]
|
||||
|
||||
top_4_labels = encoder.inverse_transform(top_4_indices)
|
||||
|
||||
print("Top 3 predicted labels:", top_4_labels)
|
||||
BIN
task-server/svm_model.sav
Normal file
BIN
task-server/svm_model.sav
Normal file
Binary file not shown.
|
|
@ -79,81 +79,82 @@
|
|||
},
|
||||
|
||||
"users" : {
|
||||
"1" : {
|
||||
"1": {
|
||||
"id": "1",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["Objective-C", "Java", "Elasticsearch"]
|
||||
},
|
||||
"2" : {
|
||||
"2": {
|
||||
"id": "2",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["SQL", "TypeScript", "jQuery"]
|
||||
},
|
||||
"3" : {
|
||||
"3": {
|
||||
"id": "3",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["Java", "Bootstrap", "jQuery"]
|
||||
},
|
||||
"4" : {
|
||||
"4": {
|
||||
"id": "4",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["HTML/CSS", "Adobe Creative Suite", "C++"]
|
||||
},
|
||||
"5" : {
|
||||
"5": {
|
||||
"id": "5",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["SQL", "PostgreSQL", "Java"]
|
||||
},
|
||||
"6" : {
|
||||
"6": {
|
||||
"id": "6",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["ReactJS", "JavaScript", "Bootstrap"]
|
||||
},
|
||||
"7" : {
|
||||
"7": {
|
||||
"id": "7",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["Ruby", "ReactJS", "Firebase"]
|
||||
},
|
||||
"8" : {
|
||||
"8": {
|
||||
"id": "8",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["Node.js", "JavaScript", "RESTful APIs"]
|
||||
},
|
||||
"9" : {
|
||||
"9": {
|
||||
"id": "9",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["HTML/CSS", "jQuery", "Go"]
|
||||
},
|
||||
"10" : {
|
||||
"10": {
|
||||
"id": "10",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["JavaScript", "ReactJS", "C++"]
|
||||
},
|
||||
"11" : {
|
||||
"11": {
|
||||
"id": "11",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": [".NET", "API Integration", "PHP"]
|
||||
},
|
||||
"12" : {
|
||||
"12": {
|
||||
"id": "12",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["SQL", "MongoDB", "Azure"]
|
||||
},
|
||||
"13" : {
|
||||
"13": {
|
||||
"id": "13",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["API Integration", "RESTful APIs", "Linux/Unix"]
|
||||
},
|
||||
"14" : {
|
||||
"14": {
|
||||
"id": "14",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["Ruby", "ReactJS", "C++"]
|
||||
},
|
||||
"15" : {
|
||||
"15": {
|
||||
"id": "15",
|
||||
"currentTasks": 0,
|
||||
"strengths" : []
|
||||
"strengths": ["JavaScript", "ReactJS", "Bootstrap"]
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
]
|
||||
Loading…
Reference in New Issue
Block a user