Select Page

Recommender Systems

Improve user experience with sophisticated search and recommendations

According to Harvard Business Review, recommender systems are the single most important distinction between “born digital” companies and legacy companies. While most organizations produce large amounts of data, very few are able to use that information to generate valuable insights. This is where recommender systems come into play, as they can provide the most relevant content to their users.

Though users may be aware of a recommender in terms of an online shopping platform that can suggest items to buy based on filters you toggle or your past purchases, a semantic recommendation system takes these activities much further. PoolParty Semantic Suite has a number of tools that can be used to build sophisticated recommender systems from its conception in a taxonomy to a front-end web application with API.

This image describes how recommendation systems can be built.

The biggest challenges for organizations who are not using semantic recommender systems.

^

Limited standard search

The standard search functionalities of a website typically does not support complex queries. This means that the results that users receive are often limited based on a small range of filters or keywords that are not being enhanced with natural language processing. In the latter case, a person wishing to research and buy a Jaguar car, might be recommended articles about the animal because of language ambiguity.

^

Unreliable filtering

Though more advanced than a standard search, a non-semantic recommender system also has its drawbacks. A system that relies on content-based filtering gives users recommendations based on past interactions or content feedback; i.e., you looked up flights in Vienna, let’s recommend you other flights for Vienna. Yet another system based on collaborative filtering gives recommendations based on other user behavior; i.e. your friend on social media liked these posts, so we will recommend you the same posts. These systems can only recommend items based on direct experience with the platform, rather than smart input based on rules that have been set for machine learning.

Useful Resources

Webinar: Agile Taxonomy Management for Customer Satisfaction.

h

Read a success story about our customer who has built powerful recommender systems from taxonomies and auto-tagging.

HR Recommender: Try our free demo to see how you can “matchmake” employees.

The biggest changes organizations can leverage with PoolParty’s semantic recommender systems.

In the case of our Vienna flights, the disadvantage here is that you may have already booked a flight to Vienna, so being recommended additional flights is counterproductive. The simple content-based filtering is limited in this regard. A semantic recommender system, however, can create “pairings” based on your activity. If you search for flights to Vienna, the system will suggest hotels that you can book in the city, excursions, restaurants, etc. These queries are far more advanced and helpful to the user who is planning their trip. 

In the business world, a semantic recommender system is incredibly useful in its ability to “matchmake” people to job opportunities in the recruitment process. Using a semantic recommender system, HR can train the machine to find profiles based on required skills, locations, etc. Once the machine is told which pairings are best, the process can be automated and recommendations can be made in real-time. An employer can find their strongest candidates by using a semantic recommender system.

Experience the major benefits of recommender systems with PoolParty Semantic Suite today.

Build from scratch

In PoolParty Semantic Suite, users can build recommendations through every step via different tools. The recommender system begins with defining concepts and labels in the Taxonomy & Thesaurus Server and identifying relationships between the specified concept types in the Ontology Server. Terms and concepts can be text mined using the Entity Extractor to further enrich the content, and then all the data is stored in a knowledge graph and trained to find matches before it is put in a front-end web application. Being able to control this process every step of the way means that your recommender system is highly tailored to your needs.

t

Complex query answering

Semantic recommender systems allow users to make sophisticated matches between the concepts that have been defined. They do not focus merely on similarities between products or business items, but on background and context as well. This behavior widens the scope of what the system can recommend, regardless of its use case.

Various business applications

These advanced systems can be applied to a variety of cases, depending on an organization’s needs. Semantic recommender systems can be used to recruit employees as candidates for jobs based on profiles, to recommend suitable consultants for projects in which they would best excel, or to enrich product information with further product knowledge, and so on. The recommender system also helps customers match products on search platforms, while giving you valuable leads on complementary information from the market.