Home
Intent and Context driven personalization
This journey demonstrates a methodology to personalize search results, by identifying clear-cut affinities/preferences across various categories that the customer has ordered for in the past.
Commerce & Cognitive
Search is an integral component of most websites. Whether it be a content site or a commerce site. The search personalization capabilities available in COTS and generally implemented are still rudimentary rule based behavior and cater to a large set of users and therefore lacks hyper personalization. Through our journey you will be able to gauge the user’s context & intent and delivery an optimized, personalized search result and reduce the number of clicks for a user to get to the content or product.
This journey demonstrates a methodology to personalize search results, by identifying clear-cut affinities/preferences across various categories that the customer has ordered for in the past.
By Priya Vasudevan, Sharath Kumar RK
https://github.com/IBM/context-driven-personalization-websphere
N/A
Search is an integral component of most websites. Whether it be a content site or a commerce site. The search personalization capabilities available in COTS and generally implemented are still rudimentary rule based behavior and cater to a large set of users and therefore lacks hyper personalization. Through our journey you will be able to gauge the user’s context & intent and delivery an optimized, personalized search result and reduce the number of clicks for a user to get to the content or product.
This journey demonstrates a methodology to personalize search results, by identifying clear-cut affinities/preferences across various categories that the customer has ordered for in the past.
When the reader has completed this journey, they will understand how to develop search personalization and boost search results, in accordance with each customer’s preferences, using the IBM WebSphere Commerce and IBM PCI (Predictive Customer Intelligence).
The intended audience for this journey are architects and senior developers who want to deliver personalization to their product / content search functionality.
-
User initiates search in WCS storefront
-
User profile data is exported from WCS to a file repository
-
Order data is exported from WCS to a file repository
-
User profile data is imported into PCI from the file repository, for analysis.
-
Order data is imported into PCI from the file repository, for analysis.
-
The PCI models establish affinity for each user across various categories based on the order data from WebSphere Commerce. The User data enriched with scores and affinity attributes is churned out from PCI.
-
This user-affinity data from PCI is fed into WebSphere Commerce and used to enrich the existing search – the affinity data is used to filter the search results.
WebSphere Commerce: IBM WebSphere Commerce is an omni-channel commerce platform that gives you the customer and business insight needed to engage shoppers as individuals with personalized content and offers, deliver mobile optimized experiences and more quickly respond to market opportunities to grow your business.
Predictive Customer Intelligence: PCI is an analytical tool that helps with reaching customer objectives and improving customer experience by analyzing the available data and predicting next best actions. The analytics in PCI is driven by IBM SPSS Modeler.
Data Science: Systems and scientific methods to analyze structured and unstructured data in order to extract knowledge and insights.
Java: A secure Object-oriented programming language, used to build applications. IBM Websphere Commerce is built using Java script, Java Server Pages and Java programming language.
What if a commerce system could understand our preferences and choices and serve up different search results, based on the same? Is it fair to serve up the same search result just because the search terms are same? How can we bring in customer’s context and intent in personalizing the search result?
This is what we try to solve with this customer journey, wherein we use each customer's browsing pattern, order history to serve up a highly personalized list of products on the WCS Aurora store. We differentiate between strong biases and soft preferences using our model, for each customer. This data is then used for applying filtering and search boosting for that individual customer. This ensures that there is true personalization & no generalization of personalization.
You may use the same logic and apply to your search on a content site too. We are not replacing the Solr engine here, so there is no drastic re-platforming. Although we have implemented this specifically for personalizing search result, the same logic can be applied to site navigation, curation of recommendation, creating dynamic menus / categories for each individual customer.
View the entire Context Based personalization pattern, including demos, code and more.