alihan/FinGPT
1
0
Fork 0
mirror of https://github.com/AI4Finance-Foundation/FinGPT.git synced 2024-02-15 23:10:01 +03:00
Code Issues Packages Projects Releases Wiki Activity

extract news classification logic out of external LLMs, put into utils

Browse Source
This commit is contained in:
raphaelzhou1
2023-10-12 07:43:59 -04:00
parent aada7601d7
commit 0a69d23afd
3 changed files with 26 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
fingpt/FinGPT-RAG/multisource_retrieval/external_LLMs/external_LLMs.py
View File

@@ -37,17 +37,6 @@ def extract_classification(text, classification_prompt):
json_data = response.json()
print("json data", json_data)
classification_response = json_data["choices"][0]['message']['content'].strip()
if "Twitter" in classification_response:
classification_response = "Twitter"
elif "Seeking Alpha" in classification_response:
classification_response = "Seeking Alpha"
elif "Reuters" in classification_response:
classification_response = "Reuters"
elif "WSJ" in classification_response:
classification_response = "WSJ"
else:
classification_response = "Unknown"
print("Classification response:", classification_response)
return classification_response
except requests.exceptions.RequestException as e:

10
fingpt/FinGPT-RAG/multisource_retrieval/utils/classify_news_sources.py
View File

@@ -34,6 +34,16 @@ try:
for row_index, row in df.iloc[1:].iterrows():
target_sentence = row[sentence_column]
classification_response = external_LLMs.extract_classification(target_sentence, classification_prompt)
if "Twitter" in classification_response:
classification_response = "Twitter"
elif "Seeking Alpha" in classification_response:
classification_response = "Seeking Alpha"
elif "Reuters" in classification_response:
classification_response = "Reuters"
elif "WSJ" in classification_response:
classification_response = "WSJ"
else:
classification_response = "Unknown"
df.at[row_index, "classification"] = classification_response # Assign classification response to the new column
output_file_path = os.path.splitext(file_path)[0] + "_classified.csv"

20
fingpt/FinGPT-RAG/multisource_retrieval/utils/sentiment_classification_by_external_LLMs.py
View File

@@ -23,7 +23,7 @@ try:
if not sentence_column:
raise ValueError("Invalid column selection")
df["classification"] = "" # Create a new column named "classification"
df["openai_inferred_sentiment"] = "" # Create a new column named "classification"
default_classification_prompt = ". For financial statement above, determine its sentiment (based on your existing knowledge). Your answer should be either \"negative\" or \"neutral\" or \"positive\""
classification_prompt = gui.enterbox("Modify the classification prompt:", "Custom Classification Prompt",
default_classification_prompt)
@@ -31,13 +31,25 @@ try:
if not classification_prompt:
classification_prompt = default_classification_prompt
counter = 0
output_file_path = os.path.splitext(file_path)[0] + "_classified.csv"
for row_index, row in df.iloc[1:].iterrows():
target_sentence = row[sentence_column]
classification_response = external_LLMs.extract_classification(target_sentence, classification_prompt)
df.at[row_index, "classification"] = classification_response # Assign classification response to the new column
if "negative" in classification_response:
classification_response = 0
elif "positive" in classification_response:
classification_response = 1
elif "neutral" in classification_response:
classification_response = 2
df.at[row_index, "openai_inferred_sentiment"] = classification_response
counter += 1
# Save the DataFrame to a CSV file every 10 rows
if counter % 10 == 0:
df.to_csv(output_file_path, index=False)
output_file_path = os.path.splitext(file_path)[0] + "_classified.csv"
df.to_csv(output_file_path, index=False)
gui.msgbox("Classification Complete")
except Exception as e:
gui.exceptionbox(str(e))