mikedoes/epo-hackahton

epo-hackahton's Issues

Create designs for page of specific patent

[Post-poned] Images do not have the correct orientation

Some patent drawings do not have the right orientation. We would need to find a way to make them properly oriented using an algorithm.

Develop the schema of XML files for the full text dataset

Sustainability score to outside metric

Combine a variety of data to produce a metric on sustainability.
Sources can include:

• Connecting the materials mentioned to an external API
• The CPC classification
• Company CO2 score accessed via an API

Estimate Magnitude of Datasets

Should we ask organisers?

First training task: supervised prediction of 1 class with patent-BERT

Let's use one of these models:

• https://huggingface.co/anferico/bert-for-patents
• https://github.com/ec-jrc/Patents4IPPC

Let's generate a small test dataset.

Goal: supervised classification for the class "CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO WASTEWATER TREATMENT OR WASTE MANAGEMENT": Y02W

Defining "green plastics"

It turns out that we don't have a precise definition of "green plastics", but we must research and propose some.
@MikeDoes can you copy the email response?

We can create a list of potential definitions of green plastics. After, we could use an Information Retrieval engine like https://github.com/ec-jrc/Patents4IPPC (fine-tuned in our area) and ask it each of our definitions.

Combining the data folders script (python)

Compute Similarity Scores between Patents Claims with Patent Number as identifiers

Store them in a matrix so that we can use it the recommendation tab of the patent

Claims in ml_dataset contains multiple instances of the same patent

The json file in ml_datasets has different entries for the same patent. It appears that the claim-text is truncated into several different json items. We should revisit the code for extracting the claims and fix this.

Create custom mosaic homepage with patent pictures

Request API Access to Open Patent Services (OPS)