12 KiB
title, draft, short_description, description, preview_image, social_preview_image, date, author, featured, tags
| title | draft | short_description | description | preview_image | social_preview_image | date | author | featured | tags | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Building a Hotel Search App with Superlinked's Multimodal Vector Embeddings | false | Transform hotel search with multi-modal vector embeddings, combining text, price, ratings, and amenities for intelligent results. | Discover how multimodal vector search revolutionizes hotel discovery by understanding complex queries with text, numerical, and categorical data. | /blog/superlinked-multimodal-search/social_preview.png | /blog/superlinked-multimodal-search/social_preview.png | 2025-04-03T00:00:00-08:00 | Filip Makraduli, David Myriel | true |
|
The proliferation of AI has transformed how people search for products, services, and educational content. Users now expect to express their needs in natural language and receive precise, tailored results that match their intent.
Let's say you're trying to book a hotel in Paris, and you have some specific criteria:
"Affordable luxury hotels near Eiffel Tower with lots of good reviews and free parking." This isn't just a search query—it's a complex set of interrelated preferences spanning multiple data types.
In this blog, we'll show you how we built The Hotel Search Demo.
Figure 1: Vectors for this app are generated via Superlinked and stored in Qdrant.
What makes this app particularly powerful is how it breaks down your natural language query into precise parameters. As you type your question at the top, you can observe the query parameters dynamically update in the left sidebar.
In this blog, we'll show you how Qdrant and Superlinked combine textual understanding, numerical reasoning, and categorical filtering to create a seamless search experience that meets modern user expectations.
Core Components
Figure 2: in a typical search or RAG app, the embedding framework (Superlinked) combines your data and its metadata into vectors. They are ingested into a Qdrant collection and indexed.
Then, a user query is also embedded and sent as a query vector to the engine for nearest neighbour retrieval. The result is a top-k similar or exact response.
SuperLinked Framework Setup: Once you setup the Superlinked server, most of the prototype work is done right out of the sample notebook. Once ready, you can host from a GitHub repository and deploy via Actions.
Qdrant Vector Database: The easiest way to store vectors is to create a free Qdrant Cloud cluster. We have simple docs that show you how to grab the API key and upsert your new vectors and run some basic searches. For this demo, we have deployed a live Qdrant Cloud cluster.
1. Vector Spaces: The Building Blocks of Intelligent Search
At the heart of Superlinked's innovation are Spaces - specialized vector embedding environments designed for different data types. Unlike conventional approaches that force all data into a single embedding format, these spaces respect the inherent characteristics of different data types.
In our demo, four distinct spaces work together: Description, Rating, Price and Rating Count. Here is how they are defined:
# Text data is embedded using a specialized language model
description_space = sl.TextSimilaritySpace(
text=hotel_schema.description,
model=settings.text_embedder_name # all-mpnet-base-v2
)
# Numerical data uses dedicated numerical embeddings with appropriate scaling
rating_space = sl.NumberSpace(
hotel_schema.rating,
min_value=0,
max_value=10,
mode=sl.Mode.MAXIMUM # Linear scale for bounded ratings
)
price_space = sl.NumberSpace(
hotel_schema.price,
min_value=0,
max_value=1000,
mode=sl.Mode.MAXIMUM,
scale=sl.LogarithmicScale() # Log scale for prices that vary widely
)
rating_count_space = sl.NumberSpace(
hotel_schema.rating_count,
min_value=0,
max_value=22500,
mode=sl.Mode.MAXIMUM,
scale=sl.LogarithmicScale() # Log scale for wide-ranging review counts
)
What makes this powerful is that each space properly preserves the semantic relationships within its domain - all while allowing these different spaces to be combined into a cohesive search experience.
Prices are embedded to maintain their proportional relationships, Text embeddings capture semantic meanings, Ratings preserve their relative quality indicators, and the Ratings Count uses logarithmic scaling to properly weight the significance of review volume.
2. Multimodal Vector Search: The Full Picture
Traditional users see vector search as typically just text-based. Both Qdrant and Superlinked transcend this limitation by supporting a rich multimodal search environment where different data types collaborate rather than compete. For our hotel demo, this means:
- Text descriptions are embedded using state-of-the-art language models that understand semantics.
- Prices use logarithmic scaling to properly handle a wide ranges of values.
- Ratings are embedded linearly to preserve their quality indicators.
- Review counts use logarithmic scaling to account for the diminishing returns of additional reviews.
Unlike approaches that stringify all data into text before embedding (resulting in unpredictable non-monotonic relationships between numbers), or systems that maintain separate indices for different attributes, Superlinked creates a unified search space where multiple attributes can be considered simultaneously with appropriate semantic relationships preserved.
The declaration of this unified index is remarkably straightforward:
index = sl.Index(
spaces=[
description_space,
price_space,
rating_space,
rating_count_space,
],
# Additional fields for hard filtering
fields=[hotel_schema.city, hotel_schema.amenities, ...]
)
3. Intelligent Query Processing: From Natural Language to Results
The query processing system in Superlinked simplifies the way search queries are built and executed. This system allows users to interact using natural language, which is then converted into multi-dimensional vector operations, thereby moving away from rigid query structures.
The query construction in the hotel demo demonstrates this power:
query = (
sl.Query(
index,
weights={
price_space: sl.Param("price_weight", description=price_description),
rating_space: sl.Param("rating_weight", description=rating_description),
# Additional space weights...
},
)
.find(hotel_schema)
.similar(description_space.text, sl.Param("description"))
.filter(hotel_schema.city.in_(sl.Param("city")))
# Additional filters...
.with_natural_query(natural_query=sl.Param("natural_query"))
)
Breaking Down the Query
This setup enables queries like "Affordable luxury hotels near Eiffel Tower with lots of good reviews and free parking." to be automatically translated into:
- A text similarity search for "luxury" and "Eiffel Tower" concepts
- Appropriate weighting for "affordable" price (lower range)
- Hard filtering for "free parking" as an amenity
- Search for "lots" (rating count) + good reviews (rating)
Unlike systems that rely on reranking after retrieval (which can miss relevant results if the initial retrieval is too restrictive) or metadata filters (which convert fuzzy preferences like "affordable" to rigid boundaries), this approach maintains the nuance of the search throughout the entire process.
4. Hybrid Search Reimagined: Solving the Modern Search Problem
Today's search landscape is dominated by discussions of hybrid search - the combination of keyword matching for precision with vector search for semantic understanding. The hotel search demo takes this concept further by implementing a multimodal hybrid search method that spans not just text retrieval methods but entire data domains.
In the hotel search demo, we see hybrid search reimagined across multiple dimensions:
- Text hybrid search: Combining exact matching (for city names, amenity keywords) with semantic similarity (for concepts like "luxury" or "family-friendly")
- Numerical hybrid search: Blending exact range filters (minimum/maximum price) with preference-based vector similarity (for concepts like "affordable" or "high-rated")
- Categorical hybrid search: Integrating hard categorical constraints (must be in Paris) with soft preferences (prefer hotels with specific amenities)
This multi-dimensional hybrid approach solves challenges facing conventional search systems:
- Single-modal vector search fails when queries span multiple data types
- Traditional hybrid search still separates keyword and vector components, which means they have to be weighed appropriately
- Separate storage per attribute forces complex result reconciliation that loses semantic nuance
- Pure filtering approaches convert preferences into binary decisions, missing the "strength" of preference
- Re-ranking strategies may lead to weaker initial retrieval, especially with broad queries
This unified approach maintains the semantic relationships of all attributes in a multi-dimensional search space, where preferences become weights rather than filters, and where hard constraints and soft preferences seamlessly coexist in the same query.
The result is a search experience that feels intuitive and "just works" - whether users are looking for "pet-friendly boutique hotels with good reviews near the city center" or "affordable family suites with pool access in resort areas" - because the system understands both the semantics and the relationships between different attributes of what users are asking for.
As search expectations continue to evolve, this multimodal and multidimensional hybrid search approach represents not just an incremental improvement but a fundamental rethinking of how search should work.
The hotel search demo showcases this vision in action, a glimpse into a future where search understands not just the words we use, but the complex, nuanced preferences they represent.
Hosting the Demo
FILIP - WE NEED TO SHOW PEOPLE HOW TO HOST IN YOUR REPO
Watch the Video
As usual, we prepared a video recording of the talk, so you can watch it at your convenience: