Impact Graph on S2M locations

The Seats2meet Mesh graph represents the network of people and their intertwined knowledge & activities on a location. The location is represented as a node, connected to the user nodes, which are in term then connected to their knowledge & activity nodes. By taking parts of this S2M network graph, one can examine data (people, knowledge  & activities) clouds that hover over a certain geographical area, i.e. a location, a city, a country etc.

Because the graph is self-learning, dynamic and always up to date,  these S2m network clouds will show ‘trends’ appearing within a certain area.  By cleverly interpreting these local clouds, location owners, event organizers or ambassadors of these areas can develop and match activities (stakeholder connections, events, knowledge transfers) based on these trends in order to make their area more relevant. Area in this sense can represent for example a location which is tied to a location operator (the owner), an event which is tied to an event host (the owner) or an S2M ambassador which covers an entire network of locations within a city.

So how do we do this matching and what does self learning mean?

As said before, the S2M graph contains nodes that represent certain entities. Everytime an event takes place between two entities (between a user and its tags, between two tags, between a tag and a location etc) the connection (relatedness), or the path between the two entities strenghtens. An event in this case can represent an interaction between these two entities, like answering a question about a certain tag, checkin in with certain tags or connecting with people using their tags. This creates strong links between entities that are related and in which the relatedness is confirmed multiple times, and weak links between entities that are related but from which the evidence is scarce. Because we want to keep the graph up-to-date the connections are weakened over time using a decay. A decay is nothing more than a systematic reduction in strengh after a given time period. This relatedness is not necessarily a semantic one (i.e. that two tags for example are close to each other by considering their meaning) but it represents the ‘matching value’ between those entities, which can be interpreted as the value of a connection when made through the connection of those entities.

Using the connections between an event and the location, a location and a city and even a city and a country, matches can be made outside of the regular scope of a location. The algorithm that operates the S2M mesh graph calculates the most efficient route through the graph to another user i.e. with the lowest costs. When a path outside the local cluster is more efficient to take, matches outside this cluster are made. If a person that is for example not in the same location, but near, and is the most relevant for another user, a connection will be proposed even though they might not be in the same location.

Via this medium, we would like to encourage you to ask any questions you might have on a technical basis about our S2M mesh algorithm. I will try to answer them under this post and if necessary compose a new blogpost using those questions!