Zbyszko Papierski - Graphs, graphs everywhere - Trójmiasto JUG [20.10.2016]

Kontynuujemy eksperymenty i dobrą passę. Tym razem regularnie spotkanie (drugie w tym tygodniu!) - także w języku angielskim.

This is for two reasons: there are quite a few devs around not natively Polish (working for international companies like Kainos, Epam, Spartez etc). We thought it might be good for them to come.
On the other hand, many of us speak mostly English at work (or wish they had) and this might be a good opportunity for them to catch up. So why not give it a try!

First person to give it a try would be Zbyszko Papierski, from Spartez.

Now about presentation: Graphs, graphs everywhere - Lucene-powered relationship exploration

We have already seen dedicated graphs solutions - like Neo4J, but did you know that most popular open source search engines like Solr and Elasticsearch also provide a way to enhance full-text search with graph search capabilities?
Elastic provides Graph module to both Elasticsearch and Kibana with its commercial offering currently known as X-Pack. Lately, Solr 6.0 provided similar features, currently limited to the search engine itself, but with features that provide an easy integration with renderers through GraphML.

Ok, but why use graph features at all? Simply put, if there are any relations in your data (and chances are quite high), there is additional knowledge to be gained from them. Whether it’s recommending some stuff to your users, based on preferences of others, or determining which of your blog posts are often read together - graphs are a way to go. Using features provided by Elasticsearch and Solr gives us additional full-text search capabilities - chances are you already use or need those in your application.

During the workshop I'll give a short to graph capabilities of both Elasticsearch and Solr. After that, we will develop a small application which will use both a full-text search and a graph search provided by Solr. Main task will be to use graph search to implement simple recommendation system - based on collaborative filtering method. Additionally, we will visualize the data on which recommendations are based.

Lucene knowledge is not a requisite - I'll provide a short introduction to the subject.

About Zbyszko: Software Engineer addicted to a broader perspective of a software development, functional coding wannabe and data analysis enthusiast. Last 5,(6) years in Allegro, switched to Spartez to tackle Atlassian products and breathe in some seaside air.