Machine Learning

Too Many Choices

January 1, 2020
EHL.AI

A recommender systems is a valuable service that can help clients deal with information overload. A recommender system can support our clients by identifying forms and information that is outside of the client’s ability to Google search.

One of the most promising recommending technology is based on the nearest-neighbor algorithm [1] but applied to user’s ratings and provides recommendations based on how they like and dislikes as it relates to a broad user community. [2] & [3]

We have all seen these “Tip of the day”, and “Did you know” suggestions, [4] but we may not understand the relevance. These are all recommendation system at work.

There are three general types of recommender systems and they are classified according to how recommendations are made:

  • Content-based Recommendation – recommendations are made based on previous purchase or selection. The suggestion is made base on attributes.
  • Collaborative Recommendation – It recommends items to users according to the item ratings of other people who have characteristics similar to their own – “Frequently bought together” and “What other items do customers buy after viewing this item” on Amazon. The analysis is based on the users’ tastes and preferences.
  • Hybrid Recommendation ­– It is a combination of content-based and collaborative recommendations.

Researchers Adomavicius and Tuzhilin in 2005 propose this multidimensional approach to incorporate contextual information into the design of the recommender system. [5]

The future recommender systems will include more in-depth and richer contextual data and sensory data. The personalized recommendation service that I want to create will only provide information that has a contextual link between the clients and documents and clients’ ‘daily-life’ preferences and behavior.

This contextual approach combines methods from (1) human-computer interaction, (2) statistics, (3) data mining, (4) machine learning, (5) information retrieval, and (6) microlearning.

Reference: A Survey of Context-Aware Recommendation Systems

[1] https://en.wikipedia.org/wiki/Nearest_neighbour_algorithm

[2] Hill, W., Stead, L., Rosenstein, M. and Furnas, G. (1995). Recommending and evaluating choices in a virtual community of use.  

[3] Shardanand, U. and Maes, P. (1995). Social information filtering: algorithms for automating “word of mouth”.

[4] Owen, D. (1986). User-Centered System Design, New Perspectives on Human-Computer Interaction.

[5] Adomavicius, G. and Tuzhilin, A. (2005).