allenai / s2-folks

Public space for the user community of Semantic Scholar APIs to share scripts, report issues, and make suggestions.
Other
144 stars 25 forks source link

Fail to search NeurIPS papers #114

Closed xvjiarui closed 8 months ago

xvjiarui commented 1 year ago

Describe the bug Since Neural Information Processing Systems changed the name from NIPS to NeurIPS, the search on NeurIPS paper breaks before the name changing for both the venues.

To Reproduce https://api.semanticscholar.org/graph/v1/paper/search?query=A%20memory%20frontier%20for%20complex%20synapses&fields=paperId,corpusId,url,title,venue,externalIds&venue=NIPS,NeurIPS,Neural%20Information%20Processing%20Systems&limit=100&offset=0 didn't return any papers with limitation on venue. But if we remove the venue by calling https://api.semanticscholar.org/graph/v1/paper/search?query=A%20memory%20frontier%20for%20complex%20synapses&fields=paperId,corpusId,url,title,venue,externalIds&limit=100&offset=0, it returns

{
      "paperId": "75c7ec96aaa5d271d541466e5d5a7807e815b488",
      "externalIds": {
        "DBLP": "conf/nips/LahiriG13",
        "MAG": "2110563098",
        "CorpusId": 616640
      },
      "corpusId": 616640,
      "url": "https://www.semanticscholar.org/paper/75c7ec96aaa5d271d541466e5d5a7807e815b488",
      "title": "A memory frontier for complex synapses",
      "venue": "NIPS"
    },

Expected behavior The api should be able to return this paper with venue constraint.

Truly appreciate any help or suggestion.

cfiorelli commented 9 months ago

Confirming observed behavior here. Any variation containing venue=NeurIPS returns no results yet without venue or with venue=NIPS returns as expected:

{"total": 16, "offset": 0, "next": 10, "data": [{"paperId": "75c7ec96aaa5d271d541466e5d5a7807e815b488", "title": "A memory frontier for complex synapses"}, {"paperId": "6b20b975e68900bf90b97a48c0fd73c847abb780", "title": "XIth Cajal Conference: New frontiers in neuron\u2010glial plasticity in health and disease"}, {"paperId": "eeace6046fa5f86a142222562269af14b9f6b363", "title": "Complex oxides for brain-inspired computing: A review."}, {"paperId": "653e4ad2bfd2dcb9932c76362397fe6c1d5f999c", "title": "Editorial: Network physiology, insights into the brain system: 2021"}, {"paperId": "bc45992dcf526961c63e0b9060a79daac15ea29d", "title": "Frontiers in Computational Neuroscience"}, {"paperId": "847e93a1616648acea60aeb58e7f1c2f70682b74", "title": "Cellular bases of behavioral plasticity: establishing and modifying synaptic circuits in the Drosophila genetic system."}, {"paperId": "c8930ae43eb05ab2d00e1b93595be0f1b6c73180", "title": "Guest editorial: Special issue on neuroimaging"}, {"paperId": "7d717ab3031efee33ce0f81e7dd0846802457903", "title": "ALLOSTERIC REGULATION OF SYNAPTIC PROCESSES THESIS FOR DOCTORAL DEGREE ( Ph . D . ) By Ranjita"}, {"paperId": "6a094e2c55d6866485f357d0e3b761f341eef5bb", "title": "Editorial: Breaking the cycle: attacking the malaria parasite in the liver"}, {"paperId": "5c52fc5a44c5da6b65ea145f0c6944d44879df49", "title": "New Challenges and Frontiers in the Research for Neuropsychiatric Disorders"}]}

smitar commented 8 months ago

Likely an error in s2ove, venue normalization

power10dan commented 8 months ago

There is a ticket for this: https://github.com/allenai/scholar/issues/38680, and I'm working on it.

smitar commented 8 months ago

Thanks @power10dan. Closing as duplicate of https://github.com/allenai/scholar/issues/38680.

power10dan commented 8 months ago

So I tried reproducing the error. Here is the result where I specified the publicationVenue field in the API:

"paperId": "75c7ec96aaa5d271d541466e5d5a7807e815b488",
 "externalIds": {"MAG": "2110563098", "DBLP": "conf/nips/LahiriG13", "CorpusId": 616640}, 
"corpusId": 616640, 
"publicationVenue": null, 
"url": "https://www.semanticscholar.org/paper/75c7ec96aaa5d271d541466e5d5a7807e815b488",
 "title": "A memory frontier for complex synapses", 
"venue": "NIPS"}

Looks like the publicationVenue is null, which might suggest that this paper is one of the few exceptions where we do not have a normalized venue name to it. I think doing a backfill on this paper should work? I've added NIPS as an alternative name to the NeurIPS venue, so we should make this paper have a normalized venue name.

cfiorelli commented 8 months ago

@xvjiarui try again your original request should now return the expected result

power10dan commented 8 months ago

Fixed in https://github.com/allenai/scholar/issues/38680.