Despite being open for editing by anyone, Wikipedia tends to be fairly reliable for a first pass on many topics. Its openness to editors is its greatest strength, but also its greatest vulnerability. Intentionally disruptive and malicious edits can be a nuisance for unsuspecting readers who may come across nonsensical sentences inserted by bad actors. These edits can also pollute downstream platforms (such as search engines) that may rely on Wikipedia to generate short summaries of relevant information.

Reverting these edits often falls to volunteer (human) editors. The aim of our project, "WikiShield," was to produce a machine learning model designed to quickly detect vandalism edits on Wikipedia so that the effort of human volunteer editors is not wasted reverting low-effort attempts at vandalism.