This section helps you with the key terms and concepts related to the SearchAssist application.
Sources
Sources are the data repositories that feed data to a search assist application. These are further classified into:
- Content includes data obtained from external sources through web crawling or file uploads.
- FAQs a list of common questions that users ask about an organization’s products or services.
- Actions are a list of tasks that are performed by the linked virtual assistant or bot built using the Kore.ai bot builder platform.
- Structured Data is an object, a set of key-value pairs from sources like databases, connectors.
Click to learn more about Source Management.
Indexing
Indexing is the technique by which search engines process information to enable quick responses to users’ queries. The search engine retrieves search responses by searching an index instead of searching the text directly. SearchAssist’s search engine applies the inverted index process and by default, indexes all the data.
Index Configuration
Index Configuration allows you to configure the fields relevant to your business use case so that the application can manipulate documents and fields to customize the data and results. The index configuration involves the following processes:
- Define Fields as the properties from the data sources that you would want to be included or excluded from the index.
- Define Traits to help identify specific entities or attributes that can be extracted from the users’ input.
- Configure Index Pipeline Stages to transform the incoming data into optimized documents using the Workbench.
- Train the application in order to build the index based on the configurations.
Click here to learn more about Index Management.
Index Pipeline
An Index Pipeline transforms the content into a document suitable for indexing. It consists of a series of configurable index pipeline stages. Each stage performs a different transformation on the data before passing the result to the next stage in the pipeline.
Workbench
Workbench combines the key aspects of the data indexing configuration process and transforms incoming data into transformed objects for indexing. It is bundled with a Simulator that provides an interactive preview of how stage rules affect a document before it is indexed.
Click here to learn more about Workbench.
Fields Mapping
Fields are key areas identified from the data sources that can be used in creating the search index schema for easy retrieval. For example, question and answer for an FAQ source, heading, and body for Files or Web content, etc.
The Field Mapping stage in Index Pipeline is used to perform the field mapping customization in an index pipeline document to the target field.
Keyword Extraction is a technique to automatically detect important words from the text stored in a field.
Traits
Traits are specific entities, attributes, or details that the users express in their conversations. Traits present in the utterance can be used in deriving the user’s intent. You can extract traits from documents and user queries that help to increase the relevance score.
Entity Extraction
In general, Entity Extraction is to extract the required information to accomplish the task. In SearchAssist, the platform uses NLP techniques to find named entities (required information) from the source field.
Semantic Meaning
Semantic analysis is the technique to understand the meaning and interpretation of words, signs, and sentence structure.
Search Configuration
Search Configuration allows you to tune the results that are displayed to the user for specific queries. This helps in improving the user experience by allowing them to view the most relevant and useful results.
Click here to learn more about Result Optimization.
Weights
Each field defined in the index configuration can be assigned a weight that plays an important role in ranking the results, these can be used to fine-tune the search relevance
Synonyms
You can ensure that the search application identifies your user queries more effectively by adding synonyms for keywords used in search queries.
Stop Words
Stop Words are a set of common words (such as the, a, to, all) that dilutes the importance of keywords in a search query. SearchAssist’s search engine is designed to ignore stop words at the query level. Ignoring these stop words from search queries before executing it helps to get more relevant results.
Facets
Facets allow users to refine their search results by selecting appropriate filters based on the field values.
Results Ranking
Results Ranking is an interactive and visual way to tune results. SearchAssist allows you to preview and customize the search results as per the business needs.
Business Rules
Business Rules allows you to define rules to personalize and fine-tune the search results based on contexts. Rules can also be used for merchandising and promotions.
Search Experience
Search experience can be customized and the way in which the results should be displayed to the search user can be configured so that the search users always see the results in a more meaningful way.
Click here to learn more about Search Experience.
Search Interface
The Search Interface is a graphical user interface used by the search users to pose their queries.
Result Templates
In SearchAssist, Result Templates decide the individual query results format in a search interface. In other terms, these templates allow you to decide how to display different types of items in a result list.
Channel
A channel enables you to configure destinations for your application to communicate with one or more channels. Click here to learn more.
Analytics
The analytics feature of the SearchAssist application provides real-time, interactive, and smart insights about the application’s operational intelligence. Learn more.
Experiments
Experiments are the best way to test multiple index and search configuration variants and pick the best one that fulfills business needs. SearchAssist allows you to create experiments with two or more variants to continuously improve search relevance and user experience. Learn more.