SCAR

What does a recommender system do?

Recommender systems have become a pervasive part of our daily online experience by analyzing past usage behavior to suggest potentially relevant content, e.g., music, movies, or books. Today, recommender systems are “the most valuable implementation of Machine Learning” according to industry experts and IT community. One prominent example is Netflix’s movie and series recommender system, which is estimated to generate over
$1 Billion revenue per year (as savings on customer acquisition).

So Why SCAR?

Since SCAR is based on Apache Solr search engine for extremely fast and efficient information retrieval, following blocks are providing its core values:

Real-Time Recommendations

Our recommender engine is “trained” live during the data import process and therefore no additional or separate training processes are needed.

Domain Agnostic

Regardless of the industry domain and related data involved, our recommender engine can support it and assist your customers with their inquiries.

Lightweight Architecture

No heavy setup required like in AWS or MS Azure due to minimalistic and very powerful recommender engine supported with Apache Solr search engine and algorithms repository.

Combinable Data & Algorithm Types

Since our recommender engine is interactions based and supports neural embeddings for any application and purpose, it is possible to include also non-standard type of data (e.g., geo-location data) and depending on the use case, implementation of various types of recommendation algorithms is possible (single or hybrid).

Customization to Business Logic

To meet the requirements of balanced, fair and unbiased recommendations our recommender engine can accept multiple business rules at once (e.g., adults only products or vice-versa, promotions of unpopular products, etc.).