created streamlit app

.gitattributes

@@ -2,6 +2,7 @@
 *.arrow filter=lfs diff=lfs merge=lfs -text
 *.bin filter=lfs diff=lfs merge=lfs -text
 *.bz2 filter=lfs diff=lfs merge=lfs -text
+*.csv filter=lfs diff=lfs merge=lfs -text
 *.ckpt filter=lfs diff=lfs merge=lfs -text
 *.ftz filter=lfs diff=lfs merge=lfs -text
 *.gz filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,65 @@
+import pandas as pd
+from sentence_transformers import SentenceTransformer
+import faiss
+import numpy as np
+import streamlit as st
+import requests
+from PIL import Image
+from io import BytesIO
+
+
+st.title('Рекомендации сериалов по описанию пользователя')
+st.divider()
+
+df = pd.read_csv('clean_series_data.csv')
+embeddings = np.load('embeddings.npy')
+
+
+def load_image_from_url(url):
+    try:
+        response = requests.get(url)
+        response.raise_for_status()
+        return Image.open(BytesIO(response.content))
+    except Exception as e:
+        st.error(f"Не удалось загрузить изображение: {e}")
+        return None
+
+
+model = SentenceTransformer('cointegrated/rubert-tiny2')
+model.cpu()
+# embeddings_desc = df['Описание'].apply(lambda x: model.encode(x))
+# embeddings_gan = df['Жанры'].apply(lambda x: model.encode(x))
+
+# embeddings = embeddings_desc + embeddings_gan
+
+embeddings = np.array(embeddings).astype(np.float32)
+faiss.normalize_L2(embeddings)
+
+dimension = embeddings.shape[1]
+index = faiss.IndexFlatIP(dimension)
+
+index.add(embeddings)
+
+query = [st.text_area('Введите описание сериала')]
+if query:
+    query_embedding = model.encode(query).astype(np.float32)
+
+    k = st.slider('Сколько сериалов рекомендовать?',
+                  min_value=1, max_value=10, value=3, step=1)
+    distances, indices = index.search(query_embedding, k)
+
+    st.subheader('Похожие сериалы:')
+    for i in range(k):
+        url = df.loc[indices[0][i]]["Изображение"]
+        image = load_image_from_url(url)
+        st.image(image)
+        st.write(f'Название: {df.loc[indices[0][i]]["Название"]}')
+        st.write(f'Рейтинг: {df.loc[indices[0][i]]["Рейтинг"]}')
+        st.write(f'Страна: {df.loc[indices[0][i]]["Страна"]}')
+        st.write(
+            f'Длительность одной серии: {df.loc[indices[0][i]]["Длительность"]}')
+        st.write(
+            f'Количество серий: {df.loc[indices[0][i]]["Количество серий"]}')
+        st.write(f'Описание: {df.loc[indices[0][i]]["Описание"]}')
+        st.write(f'Косинусное сходство: {round(distances[0][i], 2)}')
+        st.divider()

clean_series_data.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:925177133ae0f6561279290bd5bf9e34df1014d8436fb8c05e39ac047412c44a
+size 7397331

embeddings.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f005047f4c848ead774b1db9e0f3f3bc2ec18b0122b188e6770d579fdd71f0b0
+size 8696192

requirements.txt

@@ -0,0 +1,8 @@
+streamlit
+pandas
+sentence_transformers
+faiss-cpu
+numpy
+requests
+pillow
+