A Systematic Literature Review of the Use of Computational Text Analysis Methods in Intimate Partner Violence Research.

"the 21-item checklist was as follows: definition of violence discussed clearly described and motivated ipv-related hypothesis or outcome representativeness/demographics of dataset discussed and/or analysed source size and time period of dataset reported data cleaning and sampling process reported discussion of pre-processing techniques appropriate model used for hypothesis feature selection discussed and/or different features considered different models tested and compared clear and appropriate evaluation criteria evaluation outcomes reported evaluation outcomes discussed e g comparison to other work discuss misclassifications study includes discussion of model interpretability or clearly explains model rules includes ethical discussion source code and/or datasets available includes discussion of limitations of model and/or appropriate use dataset is of an appropriate size and balance of classes discussed data labelling process is explained data is labelled according to a protocol by more than one annotator and iaa reported model is tested on held-out 'test' set model is tested or deployed "in the wild" as can be seen in the prisma chart in fig 1 the search yielded 815 results of which 315 were duplicates leaving 500 unique studies.; examining aspects of the included studies offers lessons for future research particularly regarding the definition of violence open source code and overall study design.; however future work could consider making source code available for other researchers to encourage knowledge-sharing within this field."

"Declarations Conflict of InterestThe authors declare that they have no conflict of interest.     Conflict of Interest The authors declare that they have no conflict of interest.    "

"Funding Whilst conducting and at the time of submitting this research, Lilly Neubauer was funded by an EPSRC studentship in Cybersecurity, grant number EP/S022503/1. Dr Leonie Maria Tanczer’s salary was supported by a variety of research grants, including EP/S035869/1 (SPRITE+), EP/V011189/1 (REPHRAIN), ES/S004424/1 (VAMHN), MR-VO49879/1 (VISION), and MR/W009692/1 (UKRI FLF). Isabel Straw was funded by a scholarship from UK Research and Innovation, grant number EP/S021612/."