Understanding Search and Sorting in eDirectory

Introduction

When users interact with the eDirectory platform, whether by performing keyword searches or specifying a location, the system employs a sophisticated set of criteria to retrieve and order relevant results. This ensures that users can efficiently find the information they are looking for. This article delves into the specific fields considered during searches, the weighting applied to these fields, and the logic behind how search results are ultimately ordered. By understanding these mechanisms, users and administrators alike can gain valuable insights into how content is discovered and presented within the eDirectory.

Keyword Search Criteria and Weighting

When a user enters a keyword into the search bar, eDirectory examines several key fields across different content modules to identify relevant matches. These fields, along with their assigned weights that influence the relevance scoring, are outlined below:

For Listings, Events, Classifieds, and Deals:

Friendly URL: Weight 500
Title: Weight 10
Description: Weight 5
Keyword Information: Weight 1 (This includes specific keywords, category titles, and category keywords associated with the item. For Deals, the keyword information of the linked listing is also considered.)
Location Information: Weight 1 (This encompasses the address, zip code, and location name of the item.)

For Articles:

Friendly URL: Weight 200
Title: Weight 10
Abstract: Weight 5
Keyword: Weight 1

For Blog Posts:

Friendly URL: Weight 500
Title: Weight 10
Content: Weight 5
Keyword: Weight 1

The significantly high weight assigned to the Friendly URL serves a specific purpose. If a user attempts to access a detailed page of a listing that is no longer active, they are redirected to the search results page. By prioritizing the Friendly URL, the system ensures that the summary of that specific listing, if still available, appears prominently, often as the sole result.

Location-Based Search Criteria

When a user specifies a location in the "where" field, eDirectory primarily focuses on the following information associated with each item:

Location Information: This includes the item's address, zip code, and location names. All these attributes are treated with equal weight in the initial search.

Furthermore, if the "Near by" functionality is enabled by the site administrator, the search extends beyond the exact location entered. The system considers the latitude and longitude of the searched point and includes items within a defined radius. In such scenarios, the Location Information field is given a higher weight. This ensures that listings within the specified location are displayed first, followed by listings in nearby areas. For example, a search for "New York" might initially display listings within New York City, followed by relevant listings in nearby New Jersey. The ordering within the "Near by" results is primarily based on the distance of the item from the searched location.

Ordering of Search Results

The order in which search results are presented to the user is determined by a two-tiered system when searching using keywords or location:

Score: The primary sorting criterion is the "score," which is an internal relevance score calculated by the Elastic Search engine. This score reflects how closely each result matches the search terms, taking into account the weights assigned to the various fields as described earlier. A higher score indicates a greater relevance.
Item Level: In cases where multiple results have the same score, the system uses the item's level (e.g., Diamond, Gold, Silver) as a tiebreaker. Higher-level items are displayed before lower-level ones.

It is crucial to understand that while item level plays a role in the final ordering, the relevance of the search term is the paramount factor. This design principle ensures that users are presented with the most pertinent results first, even if a higher-paying "Diamond" listing has a less precise match to the search query. For instance, a Gold-level listing named "Restaurant Tokyo" will likely appear before a Diamond-level listing named "Tokyo Japanese Restaurant" when a user searches for "Restaurant Tokyo."

When users navigate through the site using widgets like "Browse by category" or "Browse by location," or when they select suggestions from the search field, the ordering logic shifts. In these instances, the system first orders results by the level of the item and then by the score.

For specific content types accessed through browsing or suggestions, additional ordering rules apply:

Articles: Ordered by the publication date (newest first).
Blog Posts: Ordered by the publication date (newest first).
Deals: Ordered by the end date of the deal (deals ending sooner appear first).
Events: First ordered by item level, and then by the event end date (events ending sooner appear first).

Finally, users are often presented with options to further refine the sorting of results based on different criteria:

Most Relevant: This applies the standard sorting logic based on score and item level (or the specific date-based ordering for articles, blogs, and deals).
Most Popular: This orders listings or items based on the number of clicks they have received, indicating their popularity among other users.

Conclusion

The search and sorting mechanisms within eDirectory are carefully designed to prioritize relevance and user experience. By assigning different weights to various content fields and employing a multi-tiered ordering system, the platform strives to deliver the most accurate and useful results for every search query and browsing interaction. Understanding these criteria provides valuable context for both users seeking information and administrators aiming to optimize the visibility of their content within the eDirectory ecosystem.