Shared Tasks @ PAN

0 already signed up

PAN fosters digital text forensics research by organizing shared task evaluations. Shared tasks are computer science events that invite researchers and practitioners to work on a specific problem of interest, the task. You are welcome to take part in any of the tasks shown below.

Authorship (11 tasks)

Author Identification

Authorship Attribution active

Given a document and a set of candidate authors, determine which of them wrote the document.

Style Change Detection active

Given a document, determine whether it was written by a single or multiple authors.

Authorship Verification

Given a pair of documents, determine whether they are written by the same author.

Author Clustering

Given a set of documents, group them by authorship.

Author Diarization

Given a document, identify and group parts that have been written by the same author.

Author Profiling

Gender Prediction active

Given a document, determine its author's gender.

Celebrity Profiling active

Given a Social Media feed, determine the owners traits.

Language Prediction

Given a document, determine its author's specific variation of their native language.

Age Prediction

Given a document, determine its author's age.

Personality Prediction

Given a document, determine its author's personality type.

Author Obfuscation

Author Masking

Given a document and a set of documents from the same author, paraphrase the former so that its author cannot be identified, anymore.

Obfuscation Evaluation

Devise and implement performance measures that quantify safeness, soundness, and/or sensibleness of an obfuscation software.

Trust (4 tasks)

Credibility Analysis

Hyperpartisan News Detection active

Given a news article, predict whether it is hyperpartisan or not.

Wikipedia Vandalism Detection

Given an edit on a Wikipedia article, determine whether it is vandalism.

Wikipedia Quality Flaw Prediction

Given a Wikipedia article, predict whether it comprises quality flaws, and if so, which.

Deception Detection

Sexual Predator Identification

Given a chat log between two people, determine whether one pretends to be a child.

Originality (6 tasks)

Text Reuse Detection (aka Plagiarism Detection)

Source Retrieval

Given a document and a search engine, retrieve likely candidates from which text might have been reused in the document, while minimizing retrieval costs.

Text Alignment

Given a pair of documents, exracts all pairs of reused passages of maximal length.

Intrinsic Plagiarism Detection

Given a document, determine if parts of it have been written by different authors.

Cross-language Text Reuse Detection

Given a document in one language and a set of documents from another, determine whether parts of the former have been reused by translation from the latter.

External Plagiarism Detection

Given a document and a set of documents, determine whether parts of the former have been reused from the latter.

Source Code Reuse Detection

Given a piece of code and code repository, determine whether parts of the former have been reused from the latter.

Propose a Task

Miss an important task for the digital text forensics? Don't hesitate to contact us and propose the task you have in mind. Organizing a task usually requires a sufficiently large dataset comprising instances of the task's underlying problem, and performance measures.

Send proposal