The Knowledge Intelligence engine scans and analyses any articles, questions, videos, PDFs, Word documents (DOCX) and PowerPoint presentations (PPTX) created or added to Fuse. The knowledge Intelligence engine then extracts important information such key phrases, and entities such as organisations, dates, people, locations and skills. By gaining a deeper understanding of the content, Fuse is able to present you with the most relevant content possible when searching.
This section includes:
Extracting keywords
Fuse’s Knowledge Intelligence engine recognises keywords and phrases that exist inside an item of content. For example, an item of content might contain the phrase 'Electric Vehicle'. Even though this phrase contains two very distinct words, Fuse is able to recognise that they are being used jointly, as well as understands the context in which the phrase is being used. These key phrases are added as additional metadata to the original content item. This means if you search for 'Articles about Electric Vehicles', any available articles containing metadata for the phrase ‘electric vehicle’ will rank higher than other articles containing the separate words 'electric' and/or 'vehicle'.
Extracting entities
Fuse is able to process the text within the content and extract entities. An entity is a piece of information that is present somewhere in the content body that matches predefined categories, such as a person, date, company or file type, which Fuse can transform into a tag. These tags are then used to help users find this content quickly in searches. These tags also allow Fuse to perform query snapping, in which Fuse automatically applies filters and facets to search results based on the entered search query.
For example, a Word document might contain the following sentence: 'John Smith's company, ACME Corp, successfully designed and produced their first electric car in 2020'. Fuse might scan this sentence within the Word document and add the following tags to make it easier to find:
- 2020 (DateTime)
- John Smith (Person)
- ACME Corp (Organisation)
Query snapping
Fuse is able to examine your search query and apply filters where available, based on what you have entered in the search bar. This is called query snapping.
Below are some examples to illustrate how query snapping works in Fuse:
Example 1
If you search for 'Health and safety event at the Excel Centre with John Smith', Fuse might intuitively apply the following facets to search results:
- Event: Filters the search results to only show events, including a list of occurrences within each event.
- Event Host: Filters to show events with a particular host, in this case a host called 'John Smith'.
- Location: Filters to only show events taking place at a specific location, in this case 'Excel Centre'.
- Tags: Applies any preexisting tags, for example, Fuse might apply tags for 'Health', 'Safety', and 'Health and Safety'.
Example 2
If you search for 'fire safety articles by Edward Francis', Fuse might intuitively apply the following facets to search results:
- Articles: Filters the search results to only show articles.
- Community: Filters to show communities that feature 'Fire Safety' in the name.
- Author: Filters to only show articles authored by 'Edward Francis'.
- Tags: Applies any preexisting tags, for example, Fuse might apply tags for 'Fire', 'Safety', and 'Fire Safety'.
Example 3
If you search for 'courses on web design', Fuse might intuitively apply the following facets to search results:
- Courses: Filters the search results to only show courses.
- Tags: Applies any preexisting tags, for example, Fuse might apply tags for 'Web', 'Design', and 'Web Design'.