INDRA (Integrated Network and Dynamical Reasoning Assembler) is an automated model assembly system interfacing with NLP systems and databases to collect knowledge, and through a process of assembly, produce causal graphs and dynamical models.
The second way to detect retractions is by looking at the CommentsCorrectionsList (example from PMID 19717156) and seeing if there is a correction with RefType=RetractionIn:
One pitfall to keep in mind is that retractions themselves also have PMIDs assigned to them. A 'retraction publication' can be recognized by again checking the CommentsCorrectionList, but this time checking for RetractionOf (examle from PMID 19718576):
<CommentsCorrectionsList>
<CommentsCorrections RefType="RetractionOf">
<RefSource>Matsui T, Suzuki S, Ujikawa K, Usui T, Gotoh S, Sugamata M, Abe S. J Med Eng Technol. 2009;33(6):481-7</RefSource>
<PMID Version="1">19484686</PMID>
</CommentsCorrections>
</CommentsCorrectionsList>
There are cases where the correction itself contains text that could become evidence in reading, so keeping track of the PMIDs of the retraction entries is also useful.
Use of retracted PMID and article type tags
Assuming the full list of retracted PMIDs is small enough, it could be stored as a resource file in INDRA and used to filter out evidence. Functionality to filter on specific article type tags can also be added if this information is extracted as well.
kkaris
commented
10 months ago
The XML publication entries of PubMed's records at https://ftp.ncbi.nlm.nih.gov/pubmed/ contain retraction records.
Retraction Entries
There are two way to collect retractions:
The first way is to use the article type tags called
PublicationType, this occurs in entries of retracted articles (example from PMID 19717156):The second way to detect retractions is by looking at the
CommentsCorrectionsList(example from PMID 19717156) and seeing if there is a correction withRefType=RetractionIn:One pitfall to keep in mind is that retractions themselves also have PMIDs assigned to them. A 'retraction publication' can be recognized by again checking the
CommentsCorrectionList, but this time checking forRetractionOf(examle from PMID 19718576):There are cases where the correction itself contains text that could become evidence in reading, so keeping track of the PMIDs of the retraction entries is also useful.
Use of retracted PMID and article type tags
Assuming the full list of retracted PMIDs is small enough, it could be stored as a resource file in INDRA and used to filter out evidence. Functionality to filter on specific article type tags can also be added if this information is extracted as well.