Squirro Glossary

Squirro Glossary#

This glossary explains Squirro-specific terminology, that is used in Squirro’s documentation.

Also see the Glossary of Industry Terms for explanations of common industry terminology.

AI Studio#: AI Studio is Squirro’s no-code AI platform. It allows users to train and deploy AI models without writing any code. Learn more at AI Studio.
Agents#: Agents combine a Large Language Model (LLM) and specialized tools to achieve a specific goal or complete a task. The agent leverages the capabilities of the LLM and enhances it by routing specific tasks to dedicated tools, which can perform functions such as content creation, analysis, enrichment, classification, fact-checking, and transformation.
Chat Queries#: Term that refers to the inputs Squirro Chat users submit. Also known as user prompts, user questions, or user queries.
Cognitive search#: Cognitive search is a the Squirro product that refers to the combination of traditional keyword search and AI technologies to provide a more intelligent search experience. Learn more at Squirro Search.
Communities#: Built on the idea of user preferences, Communities is a feature that allows end users to personalize their Squirro experience by following topics of interest. See Communities.
Community Augmentation#: Squirro feature that allows you to add additional data to a community through an external API. See Communities Augmentation.
Copilot#: Copilot refers to a Squirro Chat setting that activates the assistant mode, allowing users to receive answers in the context of the currently applied filters. See Chat Widget.
Dashboard#: The main user interface for Squirro end users. Using widgets and layers, and utilizing visibility conditions, dashboards allow developers to create complex user experiences. See Dashboards.
Entity#: An entity is a typed, named fact extracted from an Item during data processing. Entities represent structured information, such as people, organizations, locations, or products, identified within unstructured content. Each entity belongs to a single item and carries a type (for example, company), a name (for example, Squirro), and optional metadata such as a confidence score and the location within the document where it was found. If the same name appears in multiple items, each item produces its own independent entity record. Entities can be queried, filtered, and aggregated by name across an entire project, making it possible to find all items that mention a given person, organization, or product. For more information, see the Entities Widget page.
Facet#: A facet is a named property of an Item that groups it with other items sharing the same value for that property. Facets are used to filter, aggregate, and navigate the content of a project. For example, a document about a flight from London to Lagos could have a cities facet with the values London and Lagos. The Squirro user interface refers to facets as labels. All facets can be treated as labels. References to “facet” may still appear in the Squirro codebase and in some areas of the Squirro user interface, but the user interface and documentation use the term Label for all practical purposes. No migration is required. The change is a terminology update only. For more information, see the General Concepts page.
Hybrid Search#: Hybrid search refers to the combination of traditional keyword search and other search technologies, such as semantic search, using scoring profiles. Learn more at Semantic and Hybrid Search.
libNLP#: libNLP is a Squirro-specific term for the Natural Language Processing library used in Squirro. Learn more at libNLP.
LLM Prompt#: The input data provided by Squirro to the LLM includes the user’s query and system-level instructions, which the LLM uses to generate a response for the user.
Item#: An item is the fundamental unit of content in Squirro. It represents a single, independent piece of information ingested from a data source, such as a news article, a PDF document, a service ticket, an email, or a chat message. Each item carries content fields such as a title, a body, and a link to its original source, along with metadata such as a creation date, a language, Label, and Entities. For more information, see the Item Format page.
KEE#: Known Entity Extraction, commonly abbreviated as KEE, is Squirro’s technology to enrich unstructured data by linking it to structured information such as company names or products. Learn more at Known Entity Extraction.
Keyword#: “Keywords” is the field name used across the Squirro API, codebase, and data model to store structured metadata attached to an Item. See Label for the full description of this concept.
Label#: Label is the user-facing name for the structured metadata attached to an Item. Each label is a key/value pair (for example, a city label with the value London) and an item can carry multiple labels across multiple categories. Labels can be populated directly from the data source or added during data processing through enrichments. In the Squirro API and codebase, the same concept is referred to as Keyword, Tag, or Facet. For more information, see the Labels page.
Pipelet#: A pipelet is a Squirro-specific term for a custom plugin used in a Squirro pipeline. Pipelets are written in Python and can be used to enrich data, classify data, or perform other custom tasks. Learn more at Pipelets.
Pipeline#: In Squirro, the term pipeline generally refers to a project’s primary data processing pipeline which enriches, classifies, indexes, and otherwise processes ingested data. See Data Processing Pipeline.
Project Templates#: Project templates are a Squirro-specific term for a pre-configured project that can be used as a starting point for a new project. Project templates can be exported and imported within the Squirro UI to quickly duplicate, back up, or recreate projects. Learn more at Project Templates Overview.
Semantic Search#: A search technique that uses natural language processing (NLP) to understand the intent behind a query and the context in which it is being used. It is a core part of Squirro Hybrid Search. Learn more at Learn more at Semantic and Hybrid Search.
Scoring Profile#: A scoring profile is a configuration that allows you to customize the relevancy scoring of documents in Squirro. It can be applied to either all matching documents or a subset of the most relevant documents. Scoring Profiles can be used to incorporate custom scoring algorithms, boost document scores, and combine different scoring signals. They are maintained within the Configuration Service and can be configured for each project individually. To learn more, see How to Use Scoring Profiles to Customize Document Relevancy Scoring.
Sources#: In the context of Squirro Chat, the documents that answers were extracted from. The LLM indicates to Squirro which documents were used, then Squirro attempts to highlight relevant passages related to the answer. Outside of a Squirro Chat context, Squirro often refers to connected data sources simply as sources.
Source Candidates#: In the context of Squirro Chat, source candidates are a list of documents that Squirro retrieves using a combination of keyword and vector search that serve as potential reference sources for digital assistant chat responses. Up to 10 candidates are retrieved.
Squirro Chat#: Generative AI application that uses Retrieval Augmented Generation (RAG) to allow users to safely and securely chat with their own data or website content. It is offered as two products: Squirro Chat Data and Squirro Chat Web and can be embedded on any website as a digital assistant. Learn more by visiting Squirro Chat.
Tag#: A tag is a user-friendly term for a piece of structured metadata attached to an Item. It does not correspond to a specific field name in the data model or a dedicated element in the user interface. It is a conceptual term used in documentation. See also Label.
Technical Preview#: A technical preview refers to a stage of development designation for new Squirro features released for testing and evaluation purposes. Those features are not suitable for production environments. They allow users to explore and provide feedback on new capabilities before those features become generally available. Technical preview features may have certain limitations, undergo significant changes, or be subject to different support terms compared to features that are generally available. For more information about Squirro’s release process, refer to the Squirro Release Process page.