Crossing Minds API uses cutting edge machine learning techniques to analyze customer data and provide recommendations tailored for your business. Our approach builds up on two components: machine learning embeddings and improved recommendation engines. Below, we detail how, including the notions of collaborative filtering and semantic graph.

Machine Learning Embedding

Deploying complex infrastructure with heterogeneous data raises significant technical problems. From storage to real-time processing and batch analysis, all the steps of a data pipeline are made tedious when dealing with sparse and weakly structured data sets. Under the hood, our API condenses all the available data into compact, meaningful representations: machine-learning embeddings.

An embedding is a low-dimensional vector of (continuous) numbers learned to store the relevant information in a dense format. Embeddings are designed to reduce the dimensionality of both categorical and continuous features. The richness and convenience of embeddings is demonstrated by their rising role in the state-of-the-art of the machine learning literature. Thanks to their ability to represent intuitive concepts using a mathematically sound and computationally efficient structure, embeddings bring solid foundations for personalization systems.

Deep Collaborative Filtering

Deep Collaborative filtering allows to learn representations of complex patterns of human behavior in an understandable form. For this technique to reveal its potential, it is imperative to gather plenty of user data from all available sources, be it through grouping data or from referencing patterns. While simple models fail to learn simultaneously global trends and rare individual preferences from the long tail, Deep Collaborative Filtering scales to settings surpassing traditional boundaries, such as cross-domain or unbalanced data sets.

Most other solutions use “Matrix Factorization” which reduces an individual’s preferences to a simplistic linear model, thus losing what makes individuals unique. Instead, Deep Collaborative Filtering focuses on pinpointing complex patterns that our algorithms then use to determine user tastes.

Deep Content Extraction

Recommending new items is a common challenge for businesses. More generally, any item without enough history of interactions from users won’t be recommended by traditional methods. Our solution to this cold-start problem is to extract deep information from text or image data using advanced natural language processing and convolutional neural network techniques within the same deep learning model. Deep Content Extraction allows our API to recommend items no one has interacted with yet because the algorithm is able to understand the genre of the content by automatically extracting information from items such as a movie poster, synopsis, or reviews.

Other solutions commonly patch their recommender system with hand-crafted rules to overcome the cold-start problem. However this requires expensive iterations and does not scale up. Using Deep Content Extraction, our solution can generate accurate recommendations of items as soon as they are available.

Semantic Graph Embedding

Semantic Graphing makes the goal of finding correlations between seemingly unrelated data simpler by using variables such as metadata: labels, tags, genres, actors and more. By doing this we can make sense of semantic data the same way a human would. In a nutshell, if we know a user likes a certain item, and this item is connected to a second item (for instance if two movies share certain key actors), then our algorithm considers the second item as a promising candidate to recommend to this user.

Most other solutions do not include any graph databases for their recommendations. Our algorithm leverages all the available information to optimize its recommendations, enabling users to discover hidden gems.

‍