Brisbane > Connected

Team Name

Data Whisperers

Team Members

Project Description

Our project aims to harness available open and public data as much as is possible, by adapting a graph data structure and model to the existing Brisbane app. A graph enables us to connect more data to other related data. With Brisbane's disparate data sources modelled as a graph, the data itself becomes a digital representation of Brisbane itself. Brisbane, like all good modern cities, is a living breathing organism and it just makes sense to model it's data as a graph where the data model mimics real life.

Data Story

A user should be able to access all relevant open and public data via the 'Brisbane' app and on the web. On the app, a user should be able to find locations, upcoming events, choose places based on ratings as well as discover new places via a recommendation engine. Additionally, because the data is modelled as a graph, relevant data should be suggested to a user via notification (for time sensitive alerts such as rain being detected or the air quality dropping below safe levels) through to daily or weekly newsletter digests showing recommended activities upcoming over the next week.

The graph data model

A graph data model does not need to have a strict pre-defined schema or model. It can start off small and it is a more bottom-up approach and is easily adapted and expanded upon over time. This enables a Brisbane data model to evolve as Brisbane itself evolves.

A graph data model is inherently explainable and logical. A graph consists of things that are connected to other things via some relationship and to varying degrees and directions. With the open data of Brisbane modelled as a graph and grown over time, we can start to see the key nodes, the important central elements and those important connecting elements. This could be a knowledge graph of all of Brisbane. This would assist residents, business owners, visitors and city administrators.

Data driven automation

With data modelled as a large knowledge graph, and users moving within and interacting with the elements in the graph, we can make data-driven automated notifications that are pushed to users at the right place in a timely manner where users can conveniently take action. A user won't need to search for relevant information (although they can), and useful information can be surfaced and suggested to them.

Simple examples could be that rain has been detected in their proximity, to air quality dropping below safe thresholds, to road works being scheduled the next day along their route to work to suggesting a drinking water fountain when it is hot outside and a user has been on a long walk.

With our open data modelled as a graph, we can also use more sophisticated recommendation engines. Recommendation engines could recommend a cafe based on likes from other users that are like them, it may recommend a show that has a strong connection in the graph to a particular art gallery that a user frequents and has rated highly. It may recommend an event where a band is playing that people similar to them have tended to like.

We may not even be able to think of all possible insights we may be able to glean or recommend until we see how a knowledge graph of the digital Brisbane looks like. Monitoring how the graph evolves over time as Brisbane itself evolves is likely to be interesting and useful for the people living in Brisbane. We may find unexpected and surprising data interactions when we break down data silos and network (connect) or data into a knowledge graph. All of this interconnected data does not need to only add value to app users - it could also be analysed to better inform city decision-makers to best serve Brisbane.

Graph design

The actual design of the graph data model will take work, as language and ontology are likely to be different between parts of local government organisations within Brisbane, as well as between local government and residents, business owners and visitors. But it can absolutely be done. With a graph, the model can evolve over time so we would not be setting a fixed rigid schema.

Minimum viable product

The Minimum Viable Product (MVP) would consist of data already used in the current Brisbane app, enriching it with open data from the Brisbane City Experience Challenge page. The main functional difference is, the data would be modelled as a graph.

An additional feature at his MVP level would be a personalised homepage with recommended information for each individual user, based on what user data is provided. A user does not have to provide any information, however the recommendations will be modest in their value. If a user provides more information regarding likes and preferences, more personalised recommendations can be provided.

Future efforts

Once the knowledge graph concept has been built and proven its value, additional data should be added such as data from Open Street Map, live bus locations, roadwork locations, traffic congestion, train station locations and train locations and EV charging stations.

System design

The following graph shows how the knowledge graph, users and automation engine could be built.

Open API

At this point, the data in the knowledge graph should be made available as an API for use in other applications, beyond just the new updated Brisbane app and website. Any action that can be done in the app should be engineered and made available as an API for use in other applications, such as navigation, events or geo-cache applications. For non-commercial use it could be made available for free whereas for commercial use a fair fee could be charged to cover the cost of running the API service.

Broader applications

The Brisbane knowledge graph powering a new Brisbane app and website could also be used for sharing knowledge around crime, natural disasters (fire, storms and floods) and public health outbreaks. Forcing the people of Brisbane to use other sources for this information is building data silos and this should be avoided. Data regarding crime, natural disasters or public health outbreaks will be more valuable when it is enriched with the broader context contained within the whole-of-Brisbane knowledge graph.

Privacy and security

Privacy and security and controls around the data collected from users and how it is protected, anonymised, controlled and protected is hugely important, especially in open democratic cities like Brisbane. Users should not need to provide any information to access the data in the graph. However with additional data from a user, more relevant, timely and useful information can be provided to them. Clearly explaining what data from a user can provide what benefits will ensure transparency and a user will then only need to provide the personal data they want for clear benefits. Authentication and removing personally identifiable information will be a fundamental element of a modern knowledge graph for the app users. Importantly, none of a user's personal information will ever be sold or made available to any other third party.