<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Inês Soveral on Medium]]></title>
        <description><![CDATA[Stories by Inês Soveral on Medium]]></description>
        <link>https://medium.com/@inssoveral?source=rss-57fe3159acb9------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*kZ-0GhhchcmPw4hHD7YLzQ.png</url>
            <title>Stories by Inês Soveral on Medium</title>
            <link>https://medium.com/@inssoveral?source=rss-57fe3159acb9------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 07:34:22 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@inssoveral/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Hybrid Search — Where Keywords Meet Vectors, Enabling Classifieds Discovery]]></title>
            <link>https://tech.olx.com/hybrid-search-where-keywords-meet-vectors-enabling-classifieds-discovery-b7c383fe4fc4?source=rss-57fe3159acb9------2</link>
            <guid isPermaLink="false">https://medium.com/p/b7c383fe4fc4</guid>
            <category><![CDATA[hybrid-search]]></category>
            <category><![CDATA[vector-search]]></category>
            <category><![CDATA[classifieds]]></category>
            <category><![CDATA[search-engines]]></category>
            <dc:creator><![CDATA[Inês Soveral]]></dc:creator>
            <pubDate>Tue, 09 Sep 2025 15:01:43 GMT</pubDate>
            <atom:updated>2025-09-09T15:01:43.914Z</atom:updated>
            <content:encoded><![CDATA[<h3>Hybrid Search — Where Keywords Meet Vectors, Enabling Classifieds Discovery</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Rw6cGyT6zDtEvLcjD01q5g.png" /></figure><p>In the midst of a fast-paced technological revolution, where user expectations grow increasingly sophisticated, the quality of search in the classifieds space has never been more critical. In fact, users expect the search box to understand what they are looking for and produce relevant search results; if this is not the case, they will easily move on to any competitor who provides this experience.</p><p>At OLX, our search system traditionally relied on keyword matching between user queries and ad titles and descriptions — a straightforward but rigid approach. While functional, it often led to low recall or even zero results pages (ZRPs). To mitigate this, we gradually developed an extension chain logic — which will be explained in detail in later sections — to address specific edge cases and expand recall. Over time, however, this logic became increasingly complex and difficult to improve upon, calling for a disruptive solution to further evolve our search system.</p><p>In this article, we share why and how we transitioned to Hybrid Search as a retrieval strategy for our double-sided marketplace. We walk through the key changes required to support semantic search, highlight the challenges we faced, and detail the solutions put in place to overcome them. Finally, we reflect on the tangible improvements and practical benefits observed after this shift.</p><h3>What is Hybrid Search and what value does it bring?</h3><p>Despite its lack of flexibility, keyword matching remains highly effective in e-commerce. It returns results that exactly match the user’s query terms, offering clear traceability and helping users understand why specific ads appear. This is particularly useful when users know precisely what they’re looking for, and when combined with structured filters, a feature of the OLX marketplace.</p><p>However, keyword matching falls short when queries are vague, misspelled, or phrased differently from how sellers describe their listings. In such cases, vector search shines by capturing the underlying intent of a query.</p><p>Semantic search is inherently robust to synonyms, spelling errors, and phrasing variations, even being natural language friendly. These characteristics make this approach especially valuable in large or unstructured catalogs, such as OLX, where ad content is user-generated and often inconsistent.</p><p>Finally, vector search is known to improve recall and significantly reduce ZRPs, even eliminating them when no minimum similarity threshold is enforced between query and ad embeddings.</p><p>Hybrid Search combines the precision of keyword matching with the semantic depth of vector search in an ensemble-like approach, resulting in a more robust and comprehensive recall set. By leveraging the strengths of both methods, it anchors retrieval in exact matches while expanding on meaning through semantic understanding — effectively bringing out the best of both worlds.</p><p>In the context of e-commerce, Hybrid Search is the industry standard, powering the search systems of major platforms, including <em>Amazon</em>, <em>Shopify</em>, <em>Walmart</em>, <em>Airbnb,</em> and <em>Etsy</em>. At OLX, as a classifieds platform, our goal in adopting this approach is to boost recall by surfacing relevant ads that would previously have been missed and to decrease ZRPs, increasing the share of users who engage with ad listings.</p><h3>Building embeddings</h3><p>Our existing search architecture already included keyword matching, implemented in SOLR using BM25 scoring. To move toward Hybrid Search, we focused on adding a semantic retrieval component by generating embeddings.</p><p>We evaluated multiple options for embedding models and narrowed our focus to two multilingual candidates: mCLIP and multilingual-e5-small. Multilingual support was a non-negotiable requirement, as users often search in multiple languages within each market, and having the possibility to use a single model across markets would also reduce operational complexity and maintenance overhead.</p><p>We initially considered mCLIP, a multilingual extension of OpenAI’s CLIP, which in turn is a vision–language model, including two main components: an Image encoder and a Text encoder, that learns to connect images and text in a shared embedding space. Its main innovation is that it can understand natural language prompts and relate them directly to images without requiring task-specific fine-tuning. This was a logical choice as our original motivation was to explore the possibility of including image data in the ad representation. While that idea was quickly dropped after an unsuccessful experiment, we continued evaluating mCLIP’s text encoder on its own, using it to embed ad titles. The decision to use only titles and not include ad descriptions comes from the limitation this encoder has of a fixed maximum context length of 77 tokens.</p><p>Although mCLIP is known to perform reasonably well for search, its main strength lies in image–text retrieval. This led us to consider multilingual-e5-small as a more targeted alternative, as this model had also been included in an analysis comparing pre-trained embedding models done by the recommendations team for their use-case and rendered the best results.</p><p>Unlike mCLIP, E5 models are specifically optimized for semantic search and retrieval tasks. Despite its smaller transformer architecture, which may result in slightly lower accuracy compared to larger models, the E5-small variant offers lightweight inference, making it well suited for production systems. It also supports inputs up to 512 tokens, enabling us to embed both ad titles and descriptions.</p><p>For our initial iterations, the goal was to demonstrate the value of vector search within a hybrid architecture. To accelerate development, we used both models out of the box with no additional fine-tuning on OLX-specific or domain data.</p><h3>Combining retrieval sets</h3><p>Once both keyword matching and vector search were in place, the next challenge was to effectively combine their results into a unified retrieval set. Traditionally, keyword search retrieved the top 1,000 ads based on their BM25 score, which were then passed to be re-ranked, as explained in the following section.</p><p>For Hybrid Search, we initially considered applying a similarity score threshold to vector search results, selecting only those above a certain value. The idea was to correlate similarity scores with ad relevance for each query. However, offline analysis revealed that similarity scores did not consistently separate relevant from non-relevant ads — making it infeasible to define a meaningful global threshold.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ex-acBz4awab8-xRArkfdA.png" /><figcaption>Dispersion diagram plotting vector similarity scores vs relevance</figcaption></figure><p>Given this scenario, we opted for a fixed-size allocation approach. A set number of results would be retrieved via vector search, with the remainder filled by keyword matches, ensuring the total retrieval set remained capped at 1,000 ads. We began with 35 results from semantic search, as suggested by an offline analysis balancing precision and recall, and later increased this number based on insights from online experiments.</p><h3>From Two Sources to One Ranking</h3><p>As described earlier, the combined retrieval set, from both keyword and vector search, is passed to a Learning to Rank (LTR) model for final scoring.</p><p>Our LTR model is based on XGBoost and uses a range of features derived from historical user behavior, ad characteristics, and keyword-based BM25 scores. We currently operate two iterations of this model:</p><ul><li><strong>Vanilla LTR</strong>: The initial version, trained using seller contact events, which we call successful events, as the sole target signal, meaning only interactions that led to contact are considered indicators of relevance.</li><li><strong>VAS LTR</strong>: A later iteration that balances two objectives: maximizing relevance for users and ensuring visibility for Value Added Services (VAS), ads that have paid for additional exposure.</li></ul><p>For more detailed information on the LTR models, please refer to <a href="https://medium.com/olx-engineering/from-ranknet-to-lambdamart-leveraging-xgboost-for-enhanced-ranking-models-cf21f33350fb">this</a> article.</p><p>Once vector search was introduced, we ran a detailed analysis comparing the LTR feature distributions between keyword and semantic results. We found that the existing model, trained only on keyword-based retrieval, consistently assigned lower relevance scores to vector search results, even when those results were relevant. This was because the model was blind to semantic signals, burring ads with high similarity scores if there was no keyword matching, which would be particularly problematic for VAS ads retrieved via semantic search, which suffered from poor ranking positions.</p><p>To address this, we decided to retrain the LTR model, including the vector similarity score between query-ad pairs as a feature.</p><h3>Initial Experimentation and Results</h3><p>For all initial experimentations, a smaller market was used as a low-risk way to learn and prove the value of Hybrid Search before moving on to bigger markets.</p><h4>Vector Search in Extensions</h4><p>Our first experiment with vector search was conducted in the search extensions space: a chain of fallback retrieval steps designed to address edge cases and mitigate ZRPs, as outlined in the introduction. Over time, this logic had grown increasingly complex, making further improvements difficult without a disruptive alternative.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xM987b-EAl8diljVeZPSvw.png" /><figcaption>Illustration of the original extension chain logic</figcaption></figure><p>This is where good meets great. By introducing vector search within the extension flow, independent from keyword matching and the LTR model, we can isolate and measure its standalone impact. This approach also allows us to test its effectiveness without the overhead of full integration, offering a low-risk path to prove value.</p><p>The experiment involved adding a semantic search step in the existing extension chain, after only expanding to nearby locations. This new chain link used mCLIP to embed both query and ad title, retrieving results based on vector similarity alone.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mYQJrMn8XbEssxy694VbGA.png" /><figcaption>Illustration of the semantic step added extension chain logic</figcaption></figure><p><strong>Results were overwhelmingly positive</strong>, with a more <strong>than 100% increase in successful</strong> (seller contact) <strong>events per user</strong>. The overall conclusion was that users in the experiment viewed more relevant ads, as the increase in interactions attests.</p><p>This confirmed our hypothesis: <strong>vector search can meaningfully enhance recall and relevance, particularly in scenarios where keyword matching fails</strong>. Below is an example of a query that benefited from this semantic component.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KPop_pdotOge2D7PY3UIYA.png" /></figure><h4>Hybrid Search Enablement</h4><p>Having proven the value of semantic search in the extensions layer, we moved toward implementing Hybrid Search in the main flow. This meant integrating vector search with the keyword-based retrieval set and re-ranking the combined results using our Learning to Rank (LTR) model. At the time, the model in production in this market was the Vanilla LTR, trained exclusively on keyword search data. As discussed previously, this model needed retraining to incorporate vector-specific features. However, this presented a cold start problem: there was no vector search data in the main flow to train on.</p><p>Given engineering limitations and time constraints, we opted for a pragmatic solution: introducing a linear re-ranking model to temporarily replace LTR. This approach allowed us to combine the two retrieval sets while ensuring that vector search ads had enough visibility to generate interaction data for future model training.</p><p>To minimize performance degradation, we built this linear model on top of the existing logic used before LTR — the Edismax scoring function — which linearly combined different BM25 scores with predefined weights. We preserved these weights and a recency feature, which we further boosted to protect the ranking of VAS ads, since these benefit from recency resets upon service purchases. To ensure visibility for vector results, we added the vector similarity score to the ranking function and assigned it a high weight.</p><p>We then launched an A/B test using this linear re-ranking strategy, with the sole purpose of collecting data for retraining the LTR model. The results of this experiment were never expected to be strong in terms of performance: our goal was simply to “do no harm.” While the variant underperformed relative to control, the drop was smaller than anticipated. More importantly, the experiment succeeded in producing the data we needed for retraining. An offline analysis based on this data also led to an important insight: increasing the number of vector search results from 35 to 70 improved the quality of the hybrid recall set.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gKIqlj9_6bcxnyVdfV3neA.png" /><figcaption>Flow charts illustrating the differences between control and variant for Hybrid Search enablement test</figcaption></figure><h4>Hybrid Search in the main flow</h4><p>Finally, we launched our end-goal experiment to validate Hybrid Search in the main search flow. As part of our long-term plan to adopt the VAS LTR model across all markets, we retrained this model and included it as one of the test variants. However, because the market in question had been using the Vanilla LTR up to that point, and changing the ranking model represented an extra change, we also re-trained and included it as a variant. This resulted in an A/B/C test structure, allowing us to assess how Hybrid Search performs with both ranking models.</p><p>Unfortunately, the test did not reach convergence for most metrics due to traffic being split across three variants. The results were inconsistent with expectations and with what we had observed during the extensions experiment. Further investigation revealed a bug in tracking that rendered web data unreliable.</p><p>We then re-launched the experiment, this time focusing solely on the Vanilla LTR variant to avoid further traffic fragmentation. Our primary goal was to validate the impact of Hybrid Search, not to compare ranking models, after all. To support this new experiment, we used Android data from the previous test (which was unaffected by the tracking bug) to retrain the Vanilla LTR model.</p><p>Due to the earlier tracking issue, we had to run two separate experiments:</p><ul><li><strong>One on apps</strong>, where the LTR was retrained on Android data as usual.</li><li><strong>Another on mobile web (mWeb)</strong>, where the LTR was also retrained on Android data, a deviation from standard practice, which typically relies on web data.</li></ul><p>The goals of these experiments differed: on apps, the goal was to validate how Hybrid Search interacts with the retrained LTR; on mWeb, the goal was primarily to collect interaction data on vector results for future LTR retraining.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xnkrvdyIdISa6TaPR0SYlw.png" /><figcaption>Flow charts illustrating the differences between control and variant for the Hybrid Search experiment</figcaption></figure><p>The app experiment showed promising results, with an approximately 3% increase in the share of users who completed a successful event, along with positive movement across most metrics related to organic results. Overall, <strong>search efficiency improved</strong>, as reflected by a reduction in the number of searches per session.</p><p>Even more encouraging were the results from mWeb, which showed a ~5% increase in both the share of users with successful events and in ad click rates. <strong>Results were consistently positive across all key performance indicators</strong>.</p><p>Given these strong outcomes, we decided to proceed with a full rollout of <strong>Hybrid Search on all platforms in this market</strong>. 🚀</p><h3>Where to next?</h3><p>Having demonstrated the value of Hybrid Search within the OLX marketplace, our next step is to <strong>expand and refine</strong> the solution. We aim to roll it out across all markets, starting with our largest. There, we will:</p><ul><li>Continue to iterate by fine-tuning embedding models</li><li>Experiment with more effective ways to blend vector and keyword search;</li><li>Test alternative ranking strategies to further improve relevance and user experience</li></ul><p>Hybrid Search is not a finished project, but an evolving capability — one that we believe will shape the future of discovery across OLX.</p><h3>References</h3><ol><li>Büttcher, S., Clarke, C. L. A., &amp; Cormack, G. V. (2008). <em>Hybrid search: Effectively combining keywords and semantic searches</em>. In S. Bechhofer, M. Hauswirth, J. Hoffmann, &amp; M. Koubarakis (Eds.), <em>The Semantic Web: Research and Applications. ESWC 2008. Lecture Notes in Computer Science</em>, vol 5021. Springer. <a href="https://doi.org/10.1007/978-3-540-68234-9_41">https://doi.org/10.1007/978-3-540-68234-9_41</a></li><li>Chen, Guanhua, et al. “mCLIP: Multilingual CLIP via Cross-lingual Transfer.” <em>Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL 2023)</em>, 2023, pp. 10582–10600. <a href="https://aclanthology.org/2023.acl-long.728.pdf">https://aclanthology.org/2023.acl-long.728.pdf</a></li><li>Wang, Liang, et al. “Multilingual E5 Text Embeddings: A Technical Report.” <em>arXiv preprint</em> arXiv:2402.04782 (2024). <a href="https://arxiv.org/abs/2402.04782">From fuzzy information to community detection: an approach to…</a></li></ol><h3>Acknowledgments</h3><p>A special thanks to Tiago Cabo and Mahnaz Namazizavareh for their collaboration in selecting the embedding models, Jakub Krzesłowski for his support in getting these models to production, and to the Search and Ranking teams for all their hard work.</p><p>Thank you for reading! If you enjoyed this article and want to know more about OLX and our work, please visit our <a href="https://www.olxgroup.com/">website</a> or feel free to reach out to me on <a href="https://www.linkedin.com/in/inessoveral/">LinkedIn</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b7c383fe4fc4" width="1" height="1" alt=""><hr><p><a href="https://tech.olx.com/hybrid-search-where-keywords-meet-vectors-enabling-classifieds-discovery-b7c383fe4fc4">Hybrid Search — Where Keywords Meet Vectors, Enabling Classifieds Discovery</a> was originally published in <a href="https://tech.olx.com/">OLX Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>