Liferay DXP 7.4 has released an exciting new feature in beta: Semantic Search / semantic search. This feature takes the search experience to a whole new level by better understanding the intent behind user queries, delivering more precise and relevant results.
What is semantic search and why does it matter?
Unlike conventional search, which relies on exact keyword matches, semantic search analyzes the meaning behind the terms. It uses advanced AI technologies such as machine learning and natural language processing (NLP) to capture the intent of a query. This means users get matching results even when the terms they use do not appear 1:1 in the content.
Example:
• A conventional search for “Manage existing customers” would only return content that contains this exact phrase.
• Semantic search understands that the topic could be “CRM functionality” and returns relevant content such as “Update customer profiles”, “Organize contacts” or “Customer segmentation”.
How does this work in Liferay?
Semantic search in Liferay is based on the integration of Elasticsearch and its ability to work with text embeddings. These embeddings are numerical representations of texts that capture their meaning and context. Roughly speaking, it works in the following steps:
1. Text processing and indexing:
• Content in Liferay, such as blogs, pages or documents, is analyzed and converted into text embeddings with the help of AI.
• These embeddings are stored in Elasticsearch and are available for search queries.
2. Analyzing the user query:
• A search query is also converted into a text embedding that reflects the user's intent.
3. Semantic matching:
• Elasticsearch compares the embeddings of the content with that of the search query to find the most relevant results – even if the terms do not match exactly.
Practical benefits for your Liferay environment
1. Better hit rate: Users find the content they are really looking for faster, which increases satisfaction.
2. Intelligent result ranking: Content is sorted not only by keywords but also by relevance to the user's intent.
3. Easy implementation: As a beta feature, administrators can try out this functionality without having to make far-reaching changes to their platform.
An example use case
Suppose your Liferay portal is used for a knowledge management system. An employee searches for “onboarding process”. Semantic search returns:
• Manuals and checklists on employee onboarding.
• FAQ articles about new employees.
• An internal policy titled “Induction of new employees”.
Unlike conventional search, which might only return results containing the terms “onboarding process”, semantic search expands the hits to include documents with related content.
How can you try out the feature?
1. Prerequisites:
• Liferay DXP 7.4 (U70 or later).
• Elasticsearch must be configured as the search engine.
2. Activation:
• Administrators can enable semantic search via the Liferay Control Panel. It is marked as a beta feature, which means it is still being optimized.
3. Customizations:
• Use Liferay’s content management system to optimize semantic indexes. For example, you can add relevant tags or descriptions that support the AI analysis.
