How can we help?

Home > User Guides > Authorship Investigation > Investigator > Frequently Asked Questions

Frequently Asked Questions

Does the tool detect contract cheating?

Authorship Investigation does not detect contract cheating. Instead, as the name suggests, it investigates a file’s authorship using the authorship prediction engine and then presents the evidence in the form of the Authorship Investigation Report. The report then may be used to decide whether additional investigation is required but the responsibility for the judgment of any academic misconduct lies solely in the hands of the members of the institution.

Which documents should I compare?

We recommended comparing documents which have similar subject matter or genre. It is possible to compare documents across different subjects, though this will reduce the accuracy of the report. 

 

Different subjects have different common vocabulary and often different document layouts and/or punctuation guidelines. These are some of the features that our machine learning algorithm uses when making authorship predictions.

 

The comparison documents must be authored by the same individual in order to get the most accurate prediction.

How many documents can I compare against the investigation file?

You can compare ten documents against the investigation file.

What file types are compatible with Authorship Investigation?

Authorship Investigation accepts .doc, .docx, and .pdf.

What's the maximum file size for each file uploaded?

Each file must be 100mb or less and must contain at least 2000 words of text / 500 words.

How is the prediction calculated?

Authorship prediction is calculated using our authorship prediction engine. This engine uses machine learning technology to detect stylistic differences in writing between two or more documents. It relies on many of the same features used by forensic linguists when investigating authorship claims.

How do I interpret the results?

The results in the report are made up of a style difference probability, the document metadata, and select document features used to predict the probability of authorship for a file.

 

The probability that the document in question was not written by the author of the comparison documents indicates whether further investigation is needed for the suspicious work.

 

Check out our guidance on how to interpret the report.

Last modified

Tags

This page has no custom tags.

Classifications

(not set)