Table of Contents
Gladly’s search engine powers all search experiences, including all Answers Search UI and global search:
- Search to Find Conversations and Customers
- Search and Insert Answers in Conversations Through Answers Panel
- Search Answers, Manage Variables and Placeholders
It’s designed to bring up the most relevant results for your search quickly.
Below is a simple breakdown of how it works.
How Search provides results #
When you search, Gladly looks for documents1 that match what you entered. The system ranks results based on how closely they match and how often the search terms appear.
- 1In Gladly, a document means anything with readable content that can be searched, like Customer Profiles, Answers from the knowledge base, and Conversations.
Key parts of Gladly’s Search: #
- Indexing
- Gladly tracks documents in an index, which is like a catalog of searchable content (e.g., Customer Profiles, Answers, and Conversations). This makes it faster and easier to find relevant results.
- Analyzers
- Analyzers make search results more consistent by handling slight variations. For example, searching for “This” or “this” gives the same results. Analyzers can also adjust for different languages or similar words with different endings (like “do,” “does,” and “did”).
- Mapping
- Mapping organizes how documents are stored and searched within Gladly. This helps make sure that each part of a document, like titles or content sections, is treated properly when searched.
- Tokening Search Queries
- Search queries are tokenized to allow for quick searching with higher, though not perfect, accuracy. Tokenized queries provide results as you type. For instance, when you start typing an email address, you may see the desired one in the search results. However, if you continue typing the email address, the one you wanted might disappear from the options.
- Scoring and Relevance
- Gladly decides how relevant each document is to your search based on:
- Term frequency: If a term appears often in a document, it’s likely relevant.
- Inverse document frequency (IDF): Common terms are weighted lower, while unique terms are weighted higher.
- Field-length norm: Shorter fields (like titles) with the search term are considered more relevant than longer fields with the same term.
- Gladly decides how relevant each document is to your search based on:
With these steps, Gladly’s search engine finds and presents the best matches for your search.
