Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -9,7 +9,6 @@ df = pd.read_csv("top_100_influencers_combined.csv")
|
|
| 9 |
df.fillna("", inplace=True)
|
| 10 |
|
| 11 |
# Prepare text for embedding
|
| 12 |
-
# You can include 'Reach' or other metrics as needed
|
| 13 |
profile_fields = ["Name", "Niche", "Country"]
|
| 14 |
df["profile_text"] = df[profile_fields].agg(" - ".join, axis=1)
|
| 15 |
|
|
@@ -18,7 +17,6 @@ model = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
|
|
| 18 |
influencer_embeddings = model.encode(df["profile_text"].tolist(), convert_to_tensor=True)
|
| 19 |
|
| 20 |
# Recommendation logic: find top 3 by cosine similarity
|
| 21 |
-
|
| 22 |
def recommend_influencers(brand_description):
|
| 23 |
query_embedding = model.encode(brand_description, convert_to_tensor=True)
|
| 24 |
cosine_scores = util.pytorch_cos_sim(query_embedding, influencer_embeddings)[0]
|
|
@@ -39,23 +37,33 @@ def recommend_influencers(brand_description):
|
|
| 39 |
return recs
|
| 40 |
|
| 41 |
# Format recommendations into styled HTML cards
|
| 42 |
-
|
| 43 |
def format_output(brand_input):
|
| 44 |
recs = recommend_influencers(brand_input)
|
| 45 |
html = ""
|
| 46 |
for i, rec in enumerate(recs, 1):
|
| 47 |
html += f"""
|
| 48 |
-
<div
|
| 49 |
-
<h3
|
| 50 |
-
<p
|
| 51 |
-
<p
|
| 52 |
-
<p
|
| 53 |
-
<p
|
| 54 |
-
{f"<p
|
| 55 |
</div>
|
| 56 |
"""
|
| 57 |
return html
|
| 58 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
# Build the Gradio interface
|
| 60 |
demo = gr.Interface(
|
| 61 |
fn=format_output,
|
|
@@ -67,19 +75,17 @@ demo = gr.Interface(
|
|
| 67 |
),
|
| 68 |
outputs=gr.HTML(elem_id="output-cards"),
|
| 69 |
title="InfluMatch – AI-Powered Influencer Recommender",
|
| 70 |
-
description=""
|
| 71 |
-
<div style='max-width:600px; margin-bottom:1em; color:#333;'>
|
| 72 |
-
<p><strong>InfluMatch</strong> helps you discover the ideal influencers for your brand using AI-driven similarity matching.</p>
|
| 73 |
-
<p>Simply enter a short description of your brand or campaign to get three top influencer recommendations.</p>
|
| 74 |
-
</div>
|
| 75 |
-
""",
|
| 76 |
-
theme="default",
|
| 77 |
examples=[
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
]
|
|
|
|
|
|
|
|
|
|
|
|
|
| 83 |
)
|
| 84 |
|
| 85 |
if __name__ == "__main__":
|
|
|
|
| 9 |
df.fillna("", inplace=True)
|
| 10 |
|
| 11 |
# Prepare text for embedding
|
|
|
|
| 12 |
profile_fields = ["Name", "Niche", "Country"]
|
| 13 |
df["profile_text"] = df[profile_fields].agg(" - ".join, axis=1)
|
| 14 |
|
|
|
|
| 17 |
influencer_embeddings = model.encode(df["profile_text"].tolist(), convert_to_tensor=True)
|
| 18 |
|
| 19 |
# Recommendation logic: find top 3 by cosine similarity
|
|
|
|
| 20 |
def recommend_influencers(brand_description):
|
| 21 |
query_embedding = model.encode(brand_description, convert_to_tensor=True)
|
| 22 |
cosine_scores = util.pytorch_cos_sim(query_embedding, influencer_embeddings)[0]
|
|
|
|
| 37 |
return recs
|
| 38 |
|
| 39 |
# Format recommendations into styled HTML cards
|
|
|
|
| 40 |
def format_output(brand_input):
|
| 41 |
recs = recommend_influencers(brand_input)
|
| 42 |
html = ""
|
| 43 |
for i, rec in enumerate(recs, 1):
|
| 44 |
html += f"""
|
| 45 |
+
<div class='card'>
|
| 46 |
+
<h3>{i}. {rec['Name']} <span class='platform'>({rec['Platform']})</span></h3>
|
| 47 |
+
<p><strong>Niche:</strong> {rec['Niche']}</p>
|
| 48 |
+
<p><strong>Country:</strong> {rec['Country']}</p>
|
| 49 |
+
<p><strong>Engagement Rate:</strong> {rec['ER']}</p>
|
| 50 |
+
<p><strong>Followers:</strong> {rec['Followers']}</p>
|
| 51 |
+
{f"<p><strong>Reach:</strong> {rec['Reach']}</p>" if rec['Reach'] else ""}
|
| 52 |
</div>
|
| 53 |
"""
|
| 54 |
return html
|
| 55 |
|
| 56 |
+
# Custom CSS for dark-blue, professional theme
|
| 57 |
+
custom_css = """
|
| 58 |
+
body { background-color: #0a1f44; color: #f0f0f0; font-family: 'Roboto', sans-serif; }
|
| 59 |
+
#input-box textarea { background-color: #1f305b !important; color: #ffffff; border: none; border-radius: 8px; }
|
| 60 |
+
#output-cards .card { background-color: #112857; color: #e0e0e0; border: none; border-radius: 10px; padding: 1em; margin-bottom: 1em; box-shadow: 0 4px 8px rgba(0,0,0,0.3); }
|
| 61 |
+
#output-cards h3 { margin-bottom: 0.3em; color: #ffd700; font-size: 1.3em; }
|
| 62 |
+
#output-cards .platform { font-size: 0.9em; color: #cccccc; }
|
| 63 |
+
.interface-title { color: #ffffff !important; font-size: 2em !important; }
|
| 64 |
+
.interface-description { color: #c0c0c0 !important; }
|
| 65 |
+
"""
|
| 66 |
+
|
| 67 |
# Build the Gradio interface
|
| 68 |
demo = gr.Interface(
|
| 69 |
fn=format_output,
|
|
|
|
| 75 |
),
|
| 76 |
outputs=gr.HTML(elem_id="output-cards"),
|
| 77 |
title="InfluMatch – AI-Powered Influencer Recommender",
|
| 78 |
+
description="Enter your brand or campaign description and get three top influencer matches.",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 79 |
examples=[
|
| 80 |
+
["Skincare brand for Gen Z women in the US"],
|
| 81 |
+
["High-end travel gear targeting adventure enthusiasts in Europe"],
|
| 82 |
+
["Eco-friendly fitness apparel for millennials"],
|
| 83 |
+
["Gourmet pet food brand seeking Instagram influencers in Canada"]
|
| 84 |
+
],
|
| 85 |
+
css=custom_css,
|
| 86 |
+
theme="default",
|
| 87 |
+
layout="vertical",
|
| 88 |
+
allow_flagging=False
|
| 89 |
)
|
| 90 |
|
| 91 |
if __name__ == "__main__":
|