Enhanced Search
This behavior is due to Enhanced Search and Traditional Search being powered by different search engines that are separate and behave differently.
Content Ranking
Enhanced Search: Enhanced Search is powered by Cornerstone’s newest search engine. This engine indexes and intelligently surfaces training based on several variables powered by the user’s search string. It is designed to surface relevant training expeditiously.
Traditional Search: Traditional Search is powered by Cornerstone's original search engine. This engine contains a predetermined ranking model that recommends training based on a combination of titles, descriptions, and other matches driven by the user’s search string.
Predictive search results are displayed as a user is typing a search input in both Learning Search and Traditional Search pages, and the results are displayed before the search is executed. Predictive search only uses the title and description fields to show possible results and is powered by Enhanced Search.
Enhanced Search Content Ranking and Tokenization
Enhanced Search carries its ranking technology, which may return different results from Traditional Search. If the search text is present in more than one of the considered training fields, the score is aggregated. The ordering of results is determined by the total score. The fields considered for searches depend on the pages that use Enhanced Search. Please visit each page individually in CSX Online Help for more information on relevance scoring and the considered fields.
Note: No relevancy scoring will be listed in any query that only considers one field or carries equal weighting.
For Learning Search and Global Search information, see Learning Search vs. Traditional Search.
Enhanced Search presents full-text features such as instant search, fuzzy search, text splitting, text stemming, and more. Tokenizing terms is based on whitespace and special characters. Therefore, a phrase like "Compliance 2.0" becomes searchable in terms of "Compliance," "2," and "0" for an end user.
Another example involving special characters in a search term could include dashes. For example, the search term "LMS-Compliance-2023" is broken down into "LMS," "Compliance," and "2023". In this case, any training that has either "LMS," "Compliance," or "2023" in the Title, Description, or Keyword is included in the results (See search weighting above). However, any training with an exact title match is listed at the top.
Enhanced Search will tokenize a single string up to 38 characters long when tokenizing words within a search. This means any string without spaces or certain special characters greater than 38 characters will produce inconsistent results.
Example 1: This-is-a-string-of-characters-greater-than-38 as the dashes are considered as a separator within the tokenizer.
Example 2: Thisisastringofcharactersgreaterthan38 would produce inconsistent results since there are no separators in the search query, and the object being searched is longer than 38 characters.
Exact Title Match
When a user searches for a training title using Learning Search and finds an exact title match, that training item is prioritized and returned higher in the search results than other items. Users can also receive exact title matches for training items offered in languages that are not their own by searching for the training using its default title.
Display Language and Localized Languages
Enhanced Search prioritizes training localized to the searcher’s display language when partial matches are found. When exact matches are found, Enhanced Search finds the same match, regardless of the display or localized languages for that user or training.
Exact Title Match and Display Language Use Cases
Use Case 1: Gavi is a content administrator at ACME Corp. He created an online course called "The Importance of Being Modest." Gavi's display language is set to English (US), so the default language of the training and the title of the training is set to English (US). This means that English (US) is the language in which the training is displayed to the user if the course is not localized in their language. Gavi also configures a localized title in Spanish (Spain) of "La importancia de ser modesto."
Use Case 2: Maya, a user whose display language is set to Italian (Italy), logs into the system and searches for "The Importance of Being Modest." Learning Search can find an exact title match using the default title even though the course is unavailable in Maya's display language.
Use Case 3: Henri, a user whose display language is set to Spanish (Spain), logs into the system and searches for "La importancia de ser modesto" and sees Gavi's course prioritized in search results.
Use Case 4: Manuel, a user whose display language is set to Spanish (Latin America), logs into the system and searches for “La importancia.” Since the training is not localized to Spanish (Latin America), Enhanced Search will prioritize all training that matches the partial match “La importancia” and is localized to Spanish (Latin America) before Gavi’s course is shown.
Change Tracking and Data Updates for Search Results
A newly created or updated learning object is usually updated within Enhanced Search results within 15 minutes or less, with few exceptions:
- If the learning object is included as a part of a large data ingestion or a large bulk update was completed, the time for the update to be made available within search results may depend on the amount or the size of the change.
- If the learning object is created or updated at the exact moment of an update to the data set, you can expect the learning object to be available beyond 15 minutes.
- If a new user is added to the platform, the search results are typically available for that user within 1-3 hours. When a user is edited or changes are made to a user's OU, changes may take 15 minutes or less to update. In extreme cases, like bulk updates, uploads, or large API ingests, it could take longer to process depending on how the change is made and the size of the change.