Release Summary - Oct 17, 2024 (24.20)
The following key features and improvements, along with bug fixes, have been released in Algonomy CXP products in the release version 24.20 during Oct 04, 2024 – Oct 17, 2024.
FIND DISCOVER
New Find Discover: Enhanced Product Discovery with Advanced Features
The New Find Discover is an enhanced version of the classic Discover, now integrated with the new Find stack. This release retains many of the functionalities from the classic Discover, with new features continuously being added to achieve full parity.
The New Find Discover introduces powerful capabilities to enhance product discovery, improve ranking, and provide real-time, enriched insights across both online and offline channels. Designed to be scalable, flexible, and high-performing, this update delivers a seamless experience for digital merchandisers and customers alike, addressing the limitations of the legacy Discover while leveraging the advanced features of Find.
Key Benefits of New Find Discover:
-
All Catalog Products: Includes the entire catalog, not just recommendable products.
-
Real-Time Updates: Streaming Catalog integration for live product updates.
-
Find Ecosystem: Full support for Find features, including B2B.
-
Scalability: Handles large datasets efficiently, overcoming legacy Discover limitations.
-
Future-Proof: Ensures ongoing sync between Find and Discover features.
Enhancements in this Release:
-
Ingest Global Rank and Discover Required Fields into Enrichment Service: This enhancement allows required fields such as global rank to be ingested into the enrichment service, ensuring that key product attributes are available for real-time decision-making and search optimization. (Jira: ENG-25300, ENG-28517)
-
Dynamic Global Ranking Fields: Added support for dynamic fields in the global ranking system, ensuring smooth integration with metrics and co-purchase data (Jira: ENG-29181).
-
Find Attribute Settings and Sorting: Allows sorting of querytag_name and introduces attribute settings for more precise search results (Jira: ENG-28823).
-
Support for Non-Recommendable Products: Discover jobs now handle non-recommendable products, offering greater flexibility (Jira: ENG-12348).
-
Optimized Global Rank Calculation: Migrated to Spark, enhancing performance and scalability for large data sets (Jira: ENG-25299).
-
New Discover-Specific API Endpoints: A dedicated endpoint in the search service and RRServer ensures real-time data retrieval for Find Solr (Jira: ENG-28565, ENG-28462).
-
Dynamic Schema Support in SFI: Streaming Find Indexer supports dynamic fields, optimizing ranking and enrichment processes (Jira: ENG-28822)
Enterprise Dashboard
Display Category Details in Recs Test Drive for Category Recommendations
The Display Category Details feature in Recs Test Drive allows digital merchandisers to view category information alongside product details when using a strategy that recommends categories. By including the categoryRec=true parameter in recsForPlacements calls, the category details such as image, name, and category ID are displayed in a new "Categories" tab within the results section. This enhancement provides a clearer understanding of the category recommendations that will appear on the website, with direct links to the category page in the Product Catalog for easy access.
Jira: ENG-29023
Email Recommendations Preview in Recs Test Drive
The Email Recommendations Preview feature in Recs Test Drive allows digital merchandisers to easily preview email recommendations using standard email image requests. When an email placement is selected, the "Email" channel is automatically applied, enabling the configuration of key email-specific fields such as layout, campaign, and total slots. This feature simplifies email proof validation by generating image-based recommendations, and includes a "HTML code" tab for reviewing the raw email code, complete with image and click requests.
Jira: ENG-28714
Enterprise Dashboard, Find
Support for Negative Boost Values in Boost/Bury Page
Merchandisers can now assign negative boost values to bury products that don’t match specific rules on the Boost/Bury page. Instead of fractional values, users can enter negative integers ranging from -1 to -100, effectively lowering the visibility of certain products. Positive boost values remain between 0 and 100, while the negative values handle burying. This enhancement ensures more precise control over product rankings without allowing fractional numbers.
Jira: ENG-29129
Ensemble AI
Leverage Offline Data for Enhanced Outfit Generation
Ensemble AI now integrates offline order transaction data to improve the co-purchase data used in generating collections and outfits. By incorporating both online and offline data, the model provides a more comprehensive Omnichannel score (PurchaseCPOmnichannel), resulting in more accurate and relevant outfit recommendations. The system will automatically use offline transactions, if provided, for co-occurrence scoring between products. Additionally, product-to-category scoring benefits from both online and offline transaction data, enhancing the overall quality of generated ensembles.
Jira: ENG-28109
Data Engineering, Social Proof
Social Proof API: Correct Display for No Upper Limit Thresholds
The Social Proof Output Response API now correctly handles cases where no upper limit is set for a range. Previously, the API would display a value of 0, creating an inaccurate response. With this update, when no upper limit is defined, the threshold will show only the minimum value. If both minimum and maximum values are provided, the range will continue to be displayed as expected (e.g., 10 to 100). This change ensures more accurate messaging in Social Proof output.
Jira: ENG-28956
Other Feature Enhancements
The following feature enhancements and upgrades have been made in the release version 24.20 during Oct 04, 2024 – Oct 17, 2024.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Find: Business Rule: Allow Negative Boost Value
|
Validation has been implemented in the API to ensure proper handling of boost and deboost values. If the rule type is "boost," the boost value must be positive. Conversely, if the rule type is "deboost," the boost value must be negative. This enhancement ensures that boost and deboost rules are applied correctly. |
17-Oct-24 |
|
|
Find: Convert Feed to Data Javabin/JSONL and Save to Sync Store |
The feed is now successfully converted to data in Javabin/JSONL format and saved to the sync store. This enhancement supports batch indexing in Solr 9, ensuring smoother data processing and storage. |
17-Oct-24 |
|
|
Find: Migrate Batch Solr Configuration to Solr 9 |
The batch Solr configuration has been successfully migrated to Solr 9. Language containers for batch Solr 9 are now being created as expected, ensuring compatibility with the updated Solr version. |
17-Oct-24 |
|
|
Find: Deployer Service: Create Solr 9 Indexes for Batch Clients |
The Deployer Service now reads site JSONL files and successfully creates Solr 9 indexes for batch clients. |
17-Oct-24 |
|
|
Data Engineering: Configure Retention for Snowflake Fact Tables |
A weekly retention task in Snowflake now deletes old data from large fact tables, like Merch Product Placement (3 months) and Order Metrics (13 months), to control storage costs. The task runs in under an hour. |
17-Oct-24 |
|
|
Engage, Enterprise Dashboard: Configurable Response Times Fix for Engage |
The response time fix for Engage can now be configured by site, allowing a trade-off between faster response times and using Experience Optimizer across content from all campaigns with the same priority level. By default, this option is disabled but can be enabled under the Engage configuration settings for sites that prioritize speed. |
17-Oct-24 |
|
|
Enterprise Dashboard: Allow Non-Suggested Values for Attributes in Rec Restriction and Advanced Merchandising Rules |
It is now possible to add non-suggested attribute values manually in Rec Restriction and Advanced Merchandising rules when the autocomplete API times out, addressing issues faced by sites with large catalogs like Michaels US and Worten PT. This applies to creating and editing rules, including Boost Rules and restriction rules such as Do Not Recommend and Only Recommend. |
17-Oct-24 |
|
|
Discover: Allow Non-Recommendable Products in New Discover Job |
The discover job has been updated to allow non-recommendable products, enhancing its flexibility and functionality. |
17-Oct-24 |
|
|
|
Kong Tokens Reuse |
The implementation ensures that OAuth2 tokens with no expiration are reused, maintaining a single token for each service. Cached tokens are retrieved efficiently, and token validation continues to work with both cached tokens and existing tokens from the database. Additionally, indices have been added to the service_id and credential_id columns to improve the lookup speed of access tokens. |
17-Oct-24 |
Bug and Support Fixes
The following issues have been fixed in the release version 24.20 during Oct 04, 2024 – Oct 17, 2024.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Find, rrserver: Null Pointer Exception - Item Store Runtime Catalog in RRServer |
A null pointer exception in the Item Store Runtime Catalog of RRServer was causing issues when processing product associations. This affected both Find and Recommend services, as well as the streaming recommendation sanity check. The error occurred when products with null categories were encountered. The issue has been resolved now. |
17-Oct-24 |
|
|
Enterprise Dashboard: Attribute Settings Not Populating Region Field in Product Catalog |
In the Product Catalog's product details page, the region field was not populating with the appropriate region values in the auto-suggest. This issue affected the ability to view localized attribute values by region. The list of available region IDs should now appear and update dynamically based on user input. The issue has been fixed now. |
17-Oct-24 |
|
|
Find: Query Understanding Error While Processing Querytag Operation |
An issue was encountered where the query understanding configurations were not functioning properly, resulting in an error: "Error while processing querytag operation." Despite having Query Understanding enabled and the necessary configurations set, the query tags were failing to process. |
17-Oct-24 |
|
|
Enterprise Dashboard, Social Proof: Link Start and End DateTime with Site Configuration Timezone
|
An issue was identified with social proof campaign scheduling where start and end times did not align with the site's time zone settings. This caused discrepancies in campaign timing and effectiveness. The solution links the campaign scheduling to the site's configured time zone, ensuring accurate execution. The issue has been fixed now. |
17-Oct-24 |
|
|
Enterprise Dashboard, Social Proof: Social Proof: Showing Same Message When Clicking on Recommended Products |
In LSI's single-page application, social proof messages displayed the same message for all recommended products when switching between them. This issue caused the message to persist incorrectly. The issue has been fixed now. The API for experience and targeting is called for each product switch, updating the product ID and the social proof message accordingly. |
17-Oct-24 |
|
|
Enterprise Dashboard: Configurable Strategy - Hide "Use Region as a Seed" Checkbox for "Bought Together In Order" Model |
Previously, selecting the "Use Region as a Seed" checkbox with the "Bought Together In Order" model resulted in a 400 bad request error. To prevent this issue, the checkbox is now hidden for this model. The issue has been fixed, and the checkbox is no longer visible for the "Bought Together In Order" model. |
17-Oct-24 |
|
|
Enterprise Dashboard: Handle Manual & Advanced Merchandising Rules Correctly in Recs Test Drive |
In Recs Test Drive, the issue with displaying Advanced Merchandising and Manual Recommendations rules in the Rules tab has been resolved. The Type column now correctly shows "Advanced Merchandising" or "Manual Recommendations," and links now point to the correct edit rule pages. Additionally, an edge case with empty product results has been addressed, ensuring recommendations are only displayed when relevant. |
17-Oct-24 |
|
|
Social Proof: Social Proof Thresholds Not Working as Expected |
An issue was identified where the threshold for the cart page experience was set to 2, but the message was being triggered for a single purchase. This was reproducible by adding products to the cart on Abercrombie US, with messages incorrectly showing at a threshold of 1 instead of 2. The issue has been fixed, and the thresholds are now working as expected. |
17-Oct-24 |
|
|
Engage: Issue with Product Attribute Context |
There was an issue where product attribute contexts weren't being identified correctly in Engage campaigns, despite correct configurations. This caused certain products to not display as expected. The issue has been fixed, and product contexts are now being returned as expected. |
17-Oct-24 |
|
|
Recommend: Category Recommendations Allow Duplicates When Selecting Tier |
There was an issue where category recommendations, when set to use Tier 2, returned duplicate categories for configurable strategies. This caused end users to see the same category more than once. The issue has been fixed, and category recommendations now return unique categories as expected. |
17-Oct-24 |
|
|
Engage: Personalize API tagFilter Parameter Does Not Accept "." Dot Character in Tag Value |
There was an issue where the tagFilter parameter in the Personalize API did not accept a dot character in the tag value, causing content not to be returned as expected. This has now been fixed, and the API successfully returns content when using a tag with a dot character. |
17-Oct-24 |
|
|
|
Enterprise Dashboard: Attribute Missing in the Product Catalog on the Dashboard - ALL Clients |
The Region Attribute filter in the Product Catalog Dashboard was not working correctly, causing issues for all clients. The issue has now been fixed, and the Region Attribute filter is functioning as expected. |
17-Oct-24 |
|
Science: Ensemble AI - Support for Large Style Definitions |
Ensemble AI now automatically divides large style definitions into smaller sub-styles for easier management, especially when dealing with thousands of products across multiple parts. This enhancement helps merchandisers handle complex style definitions without manual effort, merging sub-styles into a single style definition. |
17-Oct-24 |
|
|
UPS: No User Profile Data |
Users encountered an error or no data was shown on the User Profile page when user IDs had special characters requiring URL encoding. The issue has been fixed now. |
23-Oct-24 |