Removing 'Including Priors' from Export

[legendarysparky]

Hi,

tldr; Does anyone know how to export relevant articles (excluding 'including priors') via a format importable to Zotero?

I am attempting to export a review via RIS. I do not need the including priors, just the articles marked relevant. I want to import the RIS into Zotero.

I have tried excluding priors using .xlsx but then I am unable to move the xlsx into Zotero easily. I have also tried looking into editing the RIS file directly but I cannot do this easily via notepad as the N1 field does not easily suggest which are in the relevant or priors group. I cannot filter RIS variables via notepad either.

I found one article discussing a similar case but this case was solved by using the xlsx method which won't work in my case given Zotero won't accept xlsx.

[michi-zuri]

The easiest would probably importing into a temporary empty zotero group the export from asreview and the collection of priors and then to delete all duplicates (both copies). the remaining can be copied to your permanent zotero collection then. Another option would be to edit the database file results.sql in your asreview project with an SQLite client, which is a bit more technical, but very doable.

[CoderDeltaLAN]

Hi, I checked the current code and I don’t think you’re missing a setting: the export endpoint always includes priors, even though ASReview’s database API already supports priors=False. So there currently isn’t a built-in RIS export for only the records you personally screened as relevant. The clean fix would be an “Exclude priors” option in the export dialog; meanwhile, assuming the project originated from RIS, you can export the .asreview project and run this small script to create a Zotero-compatible RIS without the priors:

from tempfile import TemporaryDirectory

import asreview as asr
from asreview.data import RISWriter

with TemporaryDirectory() as tmp:
    project = asr.Project.load("review.asreview", tmp)

    with project.db as db:
        results = db.get_results_table(priors=False).set_index("record_id")
        groups = db.input[["record_id", "group_id"]].set_index("record_id")

    relevant_ids = results.index[results["label"].eq(1)]
    data = project.read_input_data()
    data = data.loc[:, ~data.columns.str.startswith("asreview_")].join(groups)
    data = data.join(results.add_prefix("asreview_"), on="group_id").loc[relevant_ids]

    RISWriter.write_data(data, "relevant_without_priors.ris")

That should produce the 737 relevant records shown outside the parentheses and avoids the XLSX-to-Zotero step entirely

[legendarysparky]

This worked great! Thank you.