CodeFest 2024 on generative AI
For the second edition of EPO CodeFest, we are excited to explore the transformative potential of generative artificial intelligence in deriving new insights from patent data.
Leveraging AI to enhance innovation and support strategic decision-making offers many significant benefits. This is an important area of focus at the EPO, as it not only enriches the utility of patent data, but also maximises user impact by accelerating the advancement of new technologies, as outlined in our Strategic Plan 2028.
Join us as we push the boundaries of what's possible with AI and patent data!
- Prize ceremony
-
The ongoing CodeFest 2024 competition has brought together coders and developers from EPO member states to propose groundbreaking generative AI applications that will transform how patent data is accessed and used. Launched on 2 July, the competition's first round concluded on 16 August.
For the second edition of the EPO’s CodeFest, coders and developers were invited to harness the transformative potential of generative AI to derive new insights from patent data. In total, 120 participants from 19 countries submitted 59 proposals, 33 of which were team efforts and 26 the work of individual contenders.
On 4 December, the EPO CodeFest prize ceremony showcased the work of the six finalists and the competition’s three winners were announced.
- Finalists
-
Out of the 50 impressive CodeFest 2024 proposals which advanced to the second round, we are proud to announce the 6 finalists selected by our jury.
Each of the finalists has demonstrated exceptional creativity and innovation, and we look forward to celebrating their contributions to the future of patent data and generative AI.
Winners
InnoVisor
The team developed a platform that leverages patent data to inspire new ideas, assist users in drafting and evaluate the feasibility of those ideas by automatically cross-referencing a patent corpus for non-overlapping subject matter.
Patent Embedders
To explore trends and predict future technology combinations, the team built a visualisation tool for patents based on embeddings. The embeddings encode the text of the patents into points in a high dimensional space, then techniques for multidimensional scaling are used to create a low dimension visual representation of patent clusters.
PatentTrends
To help users find trends in a given technological area, the team created a platform to retrieve patents using natural language queries. The semantic search and retrieval of the relevant patents is based on the embeddings and metadata generated with large language models.
Finalists
AutoFeature
For their submission, the team designed an automated feature mapping tool. A user’s claim is mapped against a set of documents and a mapping score can then be used to rank said documents, with the user interface highlighting the mapped references in the documents.
EPO Dream Team
The team created a chatbot that interprets user queries, retrieves relevant patent information and generates responses. The system allows results to be re-ranked and provides references to the retrieved content for the user queries.
INSIGHTS
The team developed a system to summarise patents using large language models and use the generated summaries to perform a semantic search of the patent corpus. The search results can be used with a chatbot to discover and understand innovations.
- Winners
-
Over the past six months, 120 participants from 19 countries participated in CodeFest 2024, submitting 59 proposals, of which 33 were team efforts and 26 the work of individual contenders. After two rounds of evaluation, our jury of experts in IT, data science and patent knowledge shortlisted six finalists. The finalists’ solutions stood out for their functionality, usability, technical merit, the creativity and originality of their generative AI application and, of course, their potential impact on innovation.
The winners were announced during the CodeFest prize ceremony on 4 December as part of the Patent Knowledge Forum.
Watch the recording of the prize ceremony
EPO CodeFest 2024 winners
Grand prize winners: Nikos Gialitsis and Sevasti Papaioannou (Greece)
The InnoVisor team developed a platform that leverages patent data to visualise patent landscapes, assist users in drafting patent applications and evaluate patentability by automatically providing a similarity score between the intended subject matter and relevant prior art documents. InnoVisor received the top prize of EUR 20 000.
Nikos Gialits is a software engineer with over 5 years of experience designing and building AI-driven applications for both EU-funded and commercial projects. Nikos actively publishes research in international conferences and holds a BSc in informatics and telecommunications and an MSc in data science and information technologies.
Sevasti Papaioannou is a software engineer with over 5 years of experience working with multilingual applications, including conversational AI systems across diverse domains and languages. Passionate about leveraging AI to solve real-world problems, she focuses on creating inclusive, innovative solutions that bridge global communication gaps.
First runners-up: Segun Aroyehun (Nigeria); Giordano De Marzo, Enrico Fenoaltea, Filippo Santoro and Andrea Tacchella (Italy)
This team, the Patent Embedders, hail from academia. They created a sophisticated visualisation tool to explore patent trends and predict future technology combinations. As the first runners-up, the Patent Embedders received EUR 10 000.
Segun Aroyehun is a postdoctoral researcher in the Social Data Science Lab at the University of Konstanz. His research focuses on deep learning and natural language processing, including the adaptation, application and understanding of large language models. He is always looking for opportunities to leverage advanced computational methods for positive impact.
Giordano De Marzo is a postdoctoral researcher and lecturer at the University of Konstanz, where he is part of the Social Data Science Lab. His current research focuses on emergent behaviours in large language models societies.
Enrico Fenoaltea is a postdoctoral researcher at Centro Ricerche Enrico Fermi (CREF) in Rome, where he works on language models and the impact of AI on socioeconomic systems.
Filippo Santoro is a PhD student at Centro Ricerche Enrico Fermi (CREF) in Rome. His research focuses on the link between language and innovation, bridging language models and polya urns.
Andrea Tacchella is lead researcher at Centro Ricerche Enrico Fermi (CREF) in Rome, where he is the head of the research program on AI and innovation. He has a background in physics and complex systems.
Second runners-up: David Burns, Stephen Hibbert, Oliver Manners and Ryan Pepper (UK)
To help users discover trends in a given technological area, the PatentTrends team created a platform to retrieve patents using natural language queries. As the second runners-up, PatentTrends received EUR 5 000.
David Burns works on copyright policy at the UK Intellectual Property Office (UKIPO), where he was formerly a patent examiner. David holds a degree in copyright from King's College London.
Stephen Hibbert works in the prototyping team at Amazon Web Services (AWS), consulting with customers to identify and collaboratively build applications on AWS, usually incorporating machine learning technologies.
Oliver Manners is a trainee patent attorney at Appleyard Lees in the UK. He obtained a PhD at the University of Manchester working on modelling metalloenzymes and ruthenium catalysis.
Ryan Pepper is a software engineer at Siemens, where he works on developing predictive maintenance solutions for the engineering industry. He completed a PhD in the simulation of nanomagnetic systems at the University of Southampton.
- The coding challenge
-
Develop a generative AI application that transforms how users access, interpret and leverage patent data to foster innovation, uncover hidden opportunities and streamline decision-making processes.
Entries can take one of several approaches, and you are encouraged to consider how your proposed application can benefit users.
These potential approaches could look like, but are not limited to:
- AI-Powered patent search engine: Develop a generative AI application that can perform intuitive and context-aware searches across patent databases, delivering highly relevant results.
- Automated patent summarisation tool: Create an AI tool that generates concise and accurate summaries of patent documents.
- Predictive analytics for patent trends: Implement a generative AI system that analyses patent filing trends to predict future technological advancements and market opportunities.
- Enhanced patent mapping and visualisation: Build a tool that uses AI to create detailed visualisations and maps of patent landscapes, highlighting relationships and gaps in technology areas.
Participants
We are calling on coders or developers aged 18 and over to submit their proposals for solutions by 16 August 2024 (deadline extended). Submissions can come from an individual person or a team of up to five people. Anyone residing in one of the EPO's member states is eligible to contribute. Teams made up of EPO colleagues, external entrants or a mixture of both are all welcome!
In order to participate you must meet all eligibility criteria included in the rules of competition.
Steps
- Submit your proposal by 16 August 2024. It should include:
- The completed form with your personal information. Please note that all team members must register individually and indicate if they are part of a team.
- An outline of your proposal for solving the coding challenge. Proposals must include an explanation of your suggested approach and indicate the data selection you intend to use, possible model implementation and model evaluation.
- During the first selection process, the jury will evaluate the proposals.
- By 23 August 2024, participants will be informed if their proposal has made it to the next round.
- Participants will work on executing their proposed solution and submit their code by 23 October 2024.
- The jury will evaluate the final solution.
- Finalists are announced and the winners will receive their prizes at the virtual prize ceremony on 4 December 2024.
Prizes
The winners of Code Fest 2024 will be eligible to win one of the following prizes:
- Grand prize winner: EUR 20 000
- 1st runner-up: EUR 10 000
- 2nd runner-up: EUR 5 000
Each prize will be accompanied by a trophy.
- Jury
-
Submissions will be evaluated by senior specialists from across the EPO working in IT, data science and AI, as well as patent information and analysis.
David Horat, Director of Patent Knowledge, Directorate-General 5 Legal and International Affairs
David is the Director of Patent Knowledge at the EPO, where he leads the delivery of data, information and knowledge products and services. He began at the EPO as a patent examiner and participated in the development of the first AI pre-search algorithm. He then headed the Prior-art Data Management department, leading and growing a department of more than 150 people in over 10 locations, delivering 1 000+ data pipelines, 2+ petabytes of data and new tools. Before joining the EPO, David was IT Director at the Oceanic Platform of the Canary Islands, where he led digital transformation. He also previously worked as a software engineer at the European Organization for Nuclear Research (CERN) and Ericsson. David holds an MSc in computer science, an LL.M. in intellectual property law and business management and has completed an executive programme in leadership.
Åsa Ribbe, Principal Director Operations, Directorate-General 1 Patent Granting Process
Åsa began working at the EPO in 1997 as an examiner in telecommunications, in the field of optical communications. Throughout her career as an examiner, team manager and director, she has worked on all parts of the patent granting procedure, from classification and search to examination and opposition. Over the years, she has been active in several IP5, CPC, Trilateral classification, search and digitalisation projects, including her latest work as Business Change Manager for the Search Programme.She holds a MSc in engineering physics from Chalmers University of Technology.
Sylvia Kok-de Vries, Director of Prospects and Studies, Directorate-General 4 Corporate Services
Sylvia joined the EPO in 2022 as the Director of Prospects and Studies. Previously, she worked for over 20 years at Statistics Netherlands, both as a manager in statistical information and data management and as Director and Deputy Principal Director. Her contributions there included making statistical information more accessible to provide insights into social issues, policy development and decision-making, as well as the set-up of the data governance model used by Dutch civil services. At the EPO she is responsible for social data analytics to facilitate decision-making and policy at all levels, both internally and externally. She is particularly interested in how AI can help drive greater impact.
Franco Mascia, Data Scientist, Directorate-General 1 Patent Granting Process and Business Information Technology
Franco began working as a patent examiner at the EPO in 2015 and moved to data science in 2020, where he has worked on neural machine translation and the EPO's first generative AI solution for the Legal Interactive Platform. Franco holds a MSc and a PhD in computer science and prior to joining the EPO he was a post-doctoral researcher at IRIDIA, the Université Libre de Bruxelles’s artificial research laboratory.
Diego Eguidazu Alonso, Chief Information Officer, Business Information Technology
Diego has served as the EPO’s Chief Information Officer since 2019. In this role he has been a key contributor to the EPO’s s digital transformation, combining his expertise in technology and leadership to steer information management in a rapidly evolving digital landscape. Before joining the EPO, he had a notable career in IT management, working in various international organisations.
Diego chairs the SACEPO Working Party on e-Patent Process, the dedicated user forum accompanying the EPO's efforts in digital transformation. His professional experience has given him a keen interest in innovation, new technologies, AI and IT sustainability. - Timeline
-
The competition has five key stages
2 July – 16 August (Registration is closed)
Entrants must submit their proposals by 16 August 2024, 23.59 hrs CEST. Proposals must include an outline and explanation of the suggested approach as well as indicate the intended data selection, the possible model implementation and model evaluation.
9 – 22 August 2024
The jury evaluates the proposals for solution to decide which move on to the next round.
23 August – 23 October 2024
The coding challenge is on!
The jury communicates the outcome of their evaluation. All successful candidates will be given access to the EPO's data resources, including datasets and services that typically require subscriptions or platforms in the beta testing phase. Access will be given to:
- the Technology Intelligence Platform, the EPO’s data analysis tool offering a JupyterLab based environment, features a powerful computing infrastructure and provides easy access to the patent data contained in PATSTAT Global, PATSTAT EP Register and EP full-text through EPO-developed libraries (in beta version).
- EPO worldwide bibliographic data,
- EPO worldwide legal event data,
- EP full-text data and
- Open Patent Services.
Consider that not all data resources will be needed to implement your solution. To submit a strong proposal based on the available data sources, look at which bulk data sets best suit your needs.
The coding challenge will run for just over eight weeks and code must be submitted by 23 October at 23.59 hrs CEST.
24 October – 15 November 2024
The jury will evaluate the submitted solutions and code to selects the finalists.
18 November 2024
Announcement of the finalists!
Finalists are invited to the EPO CodeFest prize ceremony.
4 December 2024
EPO CodeFest prize ceremony
The winners of Code Fest 2024 will be announced, and prizes awarded, during the Patent Knowledge Forum on 4 December 2024.
- Registration
-
- Registration is closed
16 August 2024, 23.59 hrs CEST - Data Protection Statement (DPS)
DPS for CodeFest 2024 - Rules of Competition (RoC)
CodeFest – Rules of competition
- Registration is closed
- Background
-
The EPO's unparalleled data resources are relied upon by a range of governmental and international organisations, such as IP Offices, the OECD, International Energy Agency (IEA), International Renewable Energy Association (IRENA) and European Commission, as well as researchers and patent information specialists. The new Technology Intelligence Platform facilitates access to the EPO's wealth of data and patent knowledge. CodeFest is a major opportunity to democratise access to these data resources and contribute to effectively leveraging patent intelligence, fostering innovation and sustainability.
Further information