Table of Contents
The Enterprise version of KonaKart contains functionality to support the generation of product recommendations using an Artificial Intelligence engine. The AI engine used is the Zucchetti AI Engine which is available as an optional extra to KonaKart.
The Enterprise version of KonaKart also contains functionality to support the generation of product recommendations using a configurable internal algorithm that does not use AI techniques. This can be used when there is no AI Engine present or there is insufficient data (at least 10 products ordered by 100 or more customers) for the AI Engine to work effectively.
The objective in both cases is to calculate the set of products most likely to be of interest to customers to improve online sales.
In order to derive the product recommendations the following steps are required to be undertaken. The steps are different depending on whether you use the Zucchetti AI service or the internal KonaKart service.
In order to derive the product recommendations using the Zucchetti AI Engine the following steps are required to be undertaken. A Quartz batch job is provided to complete all steps (except the first).
In order to derive the product recommendations using the Internal KonaKart Service the following steps are required to be undertaken. A Quartz batch job is provided to complete all steps.
In KonaKart these "Product Recommendations" are known as "Suggestions". There are two types of suggestions produced. There are suggestions for products to buy based on a previously-bought product and there are suggestions created for each individual customer.
Batch jobs are provided to export the required data from KonaKart and to import the results back into KonaKart. These can be run using the included Quartz Scheduler.
Product recommendation suggestions have version numbers which allow data to be loaded for a new version of the data at the same time as the data is being used for a previous version. At a convenient time (once the data for the new version is loaded) the "current" version can be switched over by updating the AI_SUGGESTIONS_VERSION configuration variable. This versioning feature is implemented because the complete data import can take a considerable amount of time and prevents the new data from being used before it is ready to be used..
In order to provide data specific to a customer's own requirements (relevant to their own business) it is possible to send custom field values to the AI engine.
For products, the following fields are sent: custom1, custom2, custom3, custom4 and custom5.
For customers, it is possible to map any of the Customer customer fields or any Customer Tag to the 5 custom field values sent to the AI engine. As one of the parameters to the CreateProductSuggestions batch job (see konakartadmin/WEB-INF/classes/konakart_jobs.xml) you specify the source for the 5 custom fields used for the Customer export as a comma-separated set of 5 values. Custom fields (custom1, custom2, custom3, custom4, custom5) on the Customer record are identified with "#1,#2" (for custom1, custom2 etc.). CustomerTag records are identified by the CustomerTag name.
An example is: "#1,IS_MALE,#3,FAVOURITE_COLOUR,FAVOURITE_CITY,#4