Added TopicData objects

Dockerfile

@@ -8,8 +8,7 @@ RUN apt install -y git
 
 RUN pip install gunicorn==20.1.0
 RUN pip install typing-extensions
-RUN pip install topic-wizard
-RUN pip install "turftopic>=0.13.0"
+RUN pip install "turftopic[topic-wizard]"
 
 RUN useradd -m -u 1000 user
 # Switch to the "user" user
@@ -32,6 +31,7 @@ RUN git clone https://github.com/x-tabdeveloping/topicwizard
 WORKDIR $HOME/app/topicwizard
 RUN git checkout topic-arena
 RUN cp $HOME/app/main.py $HOME/app/topicwizard/main.py
-RUN cp $HOME/app/corpus.txt $HOME/app/topicwizard/corpus.txt
+RUN cp $HOME/app/keynmf_data.joblib $HOME/app/topicwizard/keynmf_data.joblib
+RUN cp $HOME/app/top2vec_data.joblib $HOME/app/topicwizard/top2vec_data.joblib
 EXPOSE 7860
 CMD gunicorn --timeout 0 -b 0.0.0.0:7860 --workers=2 --threads=4 --worker-class=gthread main:server

keynmf_data.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08ffa99e4c964c6874e7f7ecb75dadb3caf63ea3a6f8a639c117c7df9864c07f
+size 147687747

main.py

@@ -1,14 +1,10 @@
 import dash_mantine_components as dmc
-import joblib
-import numpy as np
 from dash_extensions.enrich import (Dash, DashBlueprint, Input, Output, State,
                                     dcc, exceptions, html)
-from sentence_transformers import SentenceTransformer
-from sklearn.datasets import fetch_20newsgroups
 from topicwizard.widgets import (ConceptClusters, DocumentClusters,
                                  TopicBrowser, TopicHierarchy,
                                  create_widget_container)
-from turftopic import ClusteringTopicModel, KeyNMF
+from turftopic.data import TopicData
 
 
 def create_app(blueprint):
@@ -29,27 +25,8 @@ def create_app(blueprint):
     return app
 
 
-with open("corpus.txt") as in_file:
-    corpus = in_file.read().split("\n")
-
-print("Calculating embeddings")
-encoder = SentenceTransformer("sentence-transformers/static-retrieval-mrl-en-v1")
-embeddings = encoder.encode(corpus, show_progress_bar=True)
-
-print("Fitting keynmf")
-keynmf = KeyNMF(5, encoder=encoder, random_state=42)
-keynmf_data = keynmf.prepare_topic_data(corpus, embeddings=embeddings)
-keynmf_data.hierarchy.divide_children(5)
-
-print("Fitting top2vec")
-top2vec = ClusteringTopicModel(
-    n_reduce_to=5,
-    feature_importance="centroid",
-    encoder=encoder,
-    random_state=0,
-)
-top2vec_data = top2vec.prepare_topic_data(corpus, embeddings=embeddings)
-
+keynmf_data = TopicData.from_disk("keynmf_data.joblib")
+top2vec_data = TopicData.from_disk("top2vec_data.joblib")
 print("Building blueprints.")
 keynmf_blueprint = create_widget_container(
     [TopicBrowser(), ConceptClusters(), TopicHierarchy()],

top2vec_data.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4529b2dc32c6e1b87fc8c24367fbbc20f901b60296d28feef50dbe175913a03a
+size 177605661