ads505-app / pages /topic.py
Taylor Kirk
Fresh deployment after moving datasets to hf datahub
5d4981c
import streamlit as st
from utils.topically import make_topics
st.set_page_config(layout="wide")
DATA_OPTIONS = {
'Beauty': 'All_Beauty',
'Appliances': 'Appliances',
'Baby Products': 'Baby_Products',
'Electronics': 'Electronics',
'Health and Household': 'Health_and_Household',
'Movies and TV': 'Movies_and_TV'
}
st.markdown("# Topic Modeling")
cat = st.sidebar.selectbox(
"Choose the dataset to model",
tuple(DATA_OPTIONS.keys()),
index=None
)
column = st.sidebar.selectbox("Choose a column to model", ("Text", "Title", "Both"), index=None)
if cat and column:
category = DATA_OPTIONS[cat]
topic_pipeline, fig = make_topics(
category=category,
topic_columns=column,
lemmatize=True, # or False
n1=2,
n2=3,
n_components=5,
rating=[1, 2], # optional
helpful_vote=0, # optional
new_words=None,
n_top_words=5,
# data_dir="path/to/review_data" # optional override if needed
)
st.plotly_chart(fig, use_container_width=True, config={"scrollZoom": True})