{ "id": 503419, "title": "Lack of CPCs?", "url": "/competitions/uspto-explainable-ai/discussion/503419", "author_name": "", "post_date": "2024-05-17T11:06:10.621856400Z", "votes": 13, "comment_count": 3, "views": 0, "content": "

For many publication_numbers, cpc_codes are empty in the metadata.parquet, even though they are found here.

\n

These are examples:

\n\n

\n

@addisonhoward @sohier
\nWill this problem be fixed?

", "messages": [ { "id": "2818232", "postDate": "05/17/2024 11:06:10", "content": "

For many publication_numbers, cpc_codes are empty in the metadata.parquet, even though they are found here.

\n

These are examples:

\n\n

\n

@addisonhoward @sohier
\nWill this problem be fixed?

", "rawMarkdown": "For many publication_numbers, `cpc_codes` are empty in the metadata.parquet, even though they are found [here](https://ppubs.uspto.gov/pubwebapp/static/pages/ppubsbasic.html).\n\nThese are examples:\n- https://ppubs.uspto.gov/dirsearch-public/print/downloadPdf/PP33721\n- https://ppubs.uspto.gov/dirsearch-public/print/downloadPdf/D991616\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3962786%2Ffcfc63fb752f4528c89b2c15725706f1%2Fcpc.png?generation=1715943419431991&alt=media)\n\n@addisonhoward @sohier \nWill this problem be fixed?", "votes": null }, { "id": "2883341", "postDate": "06/21/2024 22:21:28", "content": "

Has anyone found a good solution for this? I have considered making my own dataset to correct for this but once I solve the memory problem I still cannot find a great way to download all of that information.

", "rawMarkdown": "Has anyone found a good solution for this? I have considered making my own dataset to correct for this but once I solve the memory problem I still cannot find a great way to download all of that information.", "votes": null }, { "id": "2883392", "postDate": "06/21/2024 23:18:53", "content": "

Does it really matter what the actual cpc codes are. If the provided data does not contain cpc codes for certain patent, than probably even the indexing also does not contain the cpc code for that entry. I do not know if providing the actual cpc code will help or not

", "rawMarkdown": "Does it really matter what the actual cpc codes are. If the provided data does not contain cpc codes for certain patent, than probably even the indexing also does not contain the cpc code for that entry. I do not know if providing the actual cpc code will help or not", "votes": null }, { "id": "2883698", "postDate": "06/22/2024 05:01:50", "content": "

Yes, that's right. There's no point in us modifying it ourselves. But if the host modifies the cpc in both dataset and index, participants may be able to get a higher score.

", "rawMarkdown": "Yes, that's right. There's no point in us modifying it ourselves. But if the host modifies the cpc in both dataset and index, participants may be able to get a higher score.", "votes": null } ], "comments": [ { "id": 2883341, "author_name": "firlas47", "author_url": "", "post_date": "06/21/2024 22:21:28", "content": "

Has anyone found a good solution for this? I have considered making my own dataset to correct for this but once I solve the memory problem I still cannot find a great way to download all of that information.

", "votes": null, "replies": [] }, { "id": 2883392, "author_name": "shreyas9181", "author_url": "", "post_date": "06/21/2024 23:18:53", "content": "

Does it really matter what the actual cpc codes are. If the provided data does not contain cpc codes for certain patent, than probably even the indexing also does not contain the cpc code for that entry. I do not know if providing the actual cpc code will help or not

", "votes": null, "replies": [ { "id": 2883698, "author_name": "ryotayoshinobu", "author_url": "", "post_date": "06/22/2024 05:01:50", "content": "

Yes, that's right. There's no point in us modifying it ourselves. But if the host modifies the cpc in both dataset and index, participants may be able to get a higher score.

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2818232": "For many publication_numbers, `cpc_codes` are empty in the metadata.parquet, even though they are found [here](https://ppubs.uspto.gov/pubwebapp/static/pages/ppubsbasic.html).\n\nThese are examples:\n- https://ppubs.uspto.gov/dirsearch-public/print/downloadPdf/PP33721\n- https://ppubs.uspto.gov/dirsearch-public/print/downloadPdf/D991616\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3962786%2Ffcfc63fb752f4528c89b2c15725706f1%2Fcpc.png?generation=1715943419431991&alt=media)\n\n@addisonhoward @sohier \nWill this problem be fixed?", "2883341": "Has anyone found a good solution for this? I have considered making my own dataset to correct for this but once I solve the memory problem I still cannot find a great way to download all of that information.", "2883392": "Does it really matter what the actual cpc codes are. If the provided data does not contain cpc codes for certain patent, than probably even the indexing also does not contain the cpc code for that entry. I do not know if providing the actual cpc code will help or not", "2883698": "Yes, that's right. There's no point in us modifying it ourselves. But if the host modifies the cpc in both dataset and index, participants may be able to get a higher score." }, "source": "meta" }