Exploring Azure Cognitive Search's Power


This article demonstrates the transformative prowess of Azure Cognitive Search, a technological marvel that is reshaping the way we interact with data in our data-driven world. With its advanced capabilities and features, Azure Cognitive Search is not just a search engine. It is a gateway to unlocking valuable insights from vast and diverse datasets. In this article, I will show you the features and customization of Azure Cognitive Search AI capabilities using a sample dataset of hotels.

Search functionality plays a pivotal role in modern applications, serving as a bridge between users and the external data stores that contain your un-indexed data. Its importance comes from its power to quickly help users find the right information they are looking for.

In the realm of e-commerce, search enables customers to quickly locate products amidst extensive inventories, facilitating seamless shopping experiences. In educational platforms, search empowers learners to swiftly access specific topics or resources, enhancing the effectiveness of autonomous learning. Whether it is locating critical data within an enterprise system or discovering the latest news, search mechanisms streamline the journey from query to result.

The organization of this article unfolds as follows.

  • Key features and capabilities
  • Customization and AI
  • Future of Search: AI and Beyond
  • Summary


  1. To use Azure Cognitive Search, you will need an Azure Subscription, You can get one here, or if you are a student, you can create one with Azure for students.
  2. Fundamental search terminology is crucial as it paves the way for a smoother understanding of the concepts and functionalities addressed in this article. Being familiar with terms like "query," "indexing," "relevance," and more is essential. If these concepts are new to you, it is recommended to read this article first. The article also provides instructions on how to create the resource in Azure. Follow all the steps from the Demo section. Be sure to choose "hotels-sample" instead of “real estate-us-sample” during step 6 while importing data from the provided Samples and to skip the “Add Cognitive Skills (optional) section” (i.e. do not check the checkboxes. Now, concerning the “Customize target index ” and “Create an indexer” sections, you can keep their default options.

Key features and capabilities

Azure Cognitive Search is a cloud-based service that enables you to create rich and engaging search experiences for your applications. With Azure Cognitive Search, you can.

  • Index and query data from various sources, such as Azure SQL Database, Azure Blob Storage, Azure Cosmos DB, and more.
  • Apply AI capabilities through cognitive skills, such as natural language processing, computer vision, and knowledge mining, to extract insights and enrich your data.
  • Customize and fine-tune your search relevance, ranking, and scoring using built-in or custom analyzers, tokenizers, and skillsets.

Leverage semantic search features, such as semantic ranking, captions and highlights, and answers, to provide more relevant results for your users.

Customization and AI

Azure Cognitive Search can be adjusted to fit different needs.

Consider the following example: Using the Search Explorer, which relies on the index we established, and without inputting a query, all results returned exhibit a search score of 1. To elaborate, let us imagine we initiate a search for a hotel named "Thompson House". Subsequently, when we rerun the search with the query "Thompson House,".

A video of Search Explorer results to none and Thompson House queries.

Video 1. Search Explorer.

For this query, only one document was returned—the one containing information about Thompson House assigned a search score of 7.5213.

It is important to note that there might be multiple results returned by the search, each assigned a search score based on their relevance. These results are organized in descending order according to their search scores. This is where the need for adaptation and customization arises. As we aim to identify the document that best aligns with the query, we must carefully analyze which document corresponds more closely to the search terms.

One possible adaptation would be Semantic Search, a feature that uses deep neural networks to provide relevant results based on semantics, not just lexical analysis.

Consider the query “Hello! I am looking for a hotel in Albuquerque. Can you assist me with some recommendations or provide information about available options?” This query is notably more complex compared to our previous example of simply searching for the hotel name “Thompson House”.

Search Explorer results wihout Semantic Search. Double Sanctuary got 7.40 as search score

Figure 1. Query without Semantic Search returns non-relevant information.

Search Explorer results with Semantic Search. Travel Resort got 0.98 as search score and 1.61 as search rerankerScore

Figure 2. Query with Semantic Search returns relevant information.

Examining Figure 1, we might initially perceive the obtained information as relevant due to the high search score. However, upon closer inspection of the city, it becomes evident that the suggested hotel is located in Seattle rather than Albuquerque. In contrast, when Semantic Search is activated (as depicted in Figure 2), while the search score may not be as elevated as the previous search without semantic search, an additional search rerankerscore comes into play. This subsequently leads to the retrieval of a hotel situated in Albuquerque.

To enable Semantic Search. First, select a plan.

Semantic Search plan

Figure 3. Semantic Search Plan

Then, within the Search Explorer, create a configuration, and you are ready to go by selecting a configuration for your queries.

Semantic Configuration tab within Search Explorer

Figure 4. Semantic Search Configuration.

Future of Search AI and Beyond

The evolution of search technology is on the cusp of an extraordinary transformation, propelled by the integration of artificial intelligence. As AI continues to advance, the future of search is poised to deliver an even more personalized, contextually attuned, and streamlined experience in retrieving information.

Azure Cognitive Search has also AI enrichments through cognitive skills that were not included in the previous examples. The options we skipped while importing the sample dataset. These capabilities encompass diverse realms such as natural language processing, computer vision, and knowledge mining. Through these AI-driven enhancements, search tools can extract insights and augment your data, creating a wealth of enriched content.


In this article, I discussed the transformative landscape of Azure Cognitive Search, unraveling potential applications, key features, capabilities, and customization through AI.

Starting with the creation of an Azure Cognitive Search resource and importing sample data, we navigated the mechanics of query execution and search scores, highlighting their role in refining search results. Our journey extended into the future of search, where the fusion of artificial intelligence (AI) and search technology promises to reshape information retrieval.

In a future article, I will show you how to implement these search capabilities into an application.

Similar Articles