ALI Module: Automatic

Each user can select what modules he/she wants to use. ALIAS contains three kinds of modules, depending on its level of automation and the user’s required expertise in linguistics or statistics.

  • Interactive-Examiner modules require an examiner to interact with the software; this interaction involves linguistic or statistical analysis, depending on the forensic task. The Examiner does not need to be a degreed linguist, since the linguistic or statistical analytical techniques can be learned in training.
  • Interactive-Linguist modules require a linguist to interact with the software; this interaction involves linguistic or statistical analysis, depending on the forensic task. The Linguist must be degreed and have expertise that is relevant to the forensic task. ALIAS Technology linguists are all degreed linguists with specialties in main subfields of linguistics.

In the comments on speed below, the time estimate does not include any data preparation that may be necessary in a case (e.g. converting handwritten documents to electronic format, segmenting macro-texts, bundling micro-texts, importing documents, etc.).

ALI is the nickname for the Automatic or “A” modules of ALIAS. ALI contains the following modules:

ThreatAssess: Threatening communication Assessment

Task: Is this letter a real threat, or not?

Uses: ThreatAssess has been used in real-life cases to rank communications from severe to low. ThreatAssess provides an objective, statistically-based tool for law enforcement, human resources, intelligence and security analysts.

Speed: ThreatAssess runs very quickly, in minutes.

Notes: Because ThreatAssess is part of the whole ALIAS system, documents or transcribed voice messages that the user input into the system are stored and can be used in later analyses. By accessing the stored documents, the user can later compare threats to each other and to new communications, and later perform author identification on a group of threats.

Accuracy: ThreatAssess has achieved a minimum of 90% accuracy in differentiating real threats from control documents such as simulated threats, complaints, love letters and other types.

Current Languages: English

Research-in-Progress Languages: Arabic, Spanish

 

SNARE: Suicide Notes Assessment REview

Task: Is this note, letter, blog or email a real suicide note, or not?

Uses: SNARE has been used in police and family investigations, as part of homicide investigations. SNARE provides an objective, statistically-based tool for law enforcement, human resources, psychologists, psychiatrists and other analysts.

Speed: SNARE runs very quickly, in minutes.

Notes: Law enforcement officers and medical examiners are expected to make instant decisions about suicide notes, but research shows that identifying a note as a real suicide note is extremely difficult to do, even for psychotherapists. SNARE helps law enforcement and medical examiners by providing expertise in the algorithm.

Accuracy: Litigation-independent testing has shown accuracy rates higher than the reported rates by psychiatrists and psychologists, and higher than other software as reported in the scientific literature. SNARE’s most recent validation testing achieved 88.6% accuracy, on notes shorter than 45 words.

Current Languages: English

Research-in-Progress Languages: None (we are open to requests!)

 

UniAIDE1: Unicode Author Identification Estimator

Task: Given a questioned document, how similar are documents from suspects in the pool of potential authors?

Uses: UniAIDE can help investigators determine if the actual author is or is not in the pool of potential authors. This can help to narrow or widen the search for potential authors. UniAIDE ranks documents by similarity to the questioned document, and thus identifies authors as similar to the questioned document.

Speed: UniAIDE1 runs very quickly, in minutes.

Notes: UniAIDE requires at least 5000 words per suspect author. Further, UniAIDE requires a large group of distractor documents. The automatic version allows for the user to input his/her own distractor documents or to search the ALIAS documents database (TATTLER) for gathering distractor documents. Currently, UniAIDE should not be used in court, because it is a screening device only. Its primary benefits are that it runs very quickly, processes texts in any computer-accessible language, and can eliminate some suspects.

Accuracy: UniAIDE currently ranks the actual author within the top five slots with about 80% accuracy. That is, there is about an 80% chance that the actual author will appear in one of the top five slots.

Current Languages: English, Spanish, Arabic, Chinese, Greek, German, Dutch, Italian and others

Research-in-Progress Languages: None

 

WISER1: WItness Statement Evaluation Rank

Task: Does this text contain deceptive language?

Uses: WISER can help investigators prepare for interrogations by analyzing witness statements after the interview but before the interrogation.

Speed: WISER1 runs very quickly, in minutes.

Notes: Law enforcement agencies who enter into a research relationship with the Institute for Linguistic Evidence (ALIAS Technology’s sister for R&D) can obtain access to WISER1 without cost for a negotiated period of time.

Accuracy: WISER currently attains over 90% accuracy distinguishing truthful from false witness statements from actual criminal investigations. However, the Institute for Linguistic Evidence is conducting ongoing research on new text collections to determine under what conditions WISER can continue this high level of accuracy.

Current Languages: English

Research-in-Progress Languages: Spanish

 

Chars1 and Chars2: Characters in Text(1) and Characters in Texts(2)

Task: What are the distributions of graphic characters in a text (Chars 1)? This tells are the most frequent characters and character combinations. What kind of overlaps between texts occur at the graphic character level (Chars 2)? This tells us what to focus on when comparing documents.

Uses: Chars1 and Chars2 assist forensic handwriting examiners.

Speed: Chars1 and Chars2 run very quickly, in minutes.

Notes: In forensic handwriting examination, characters and characters in stroke-groups are compared visually. Using Chars1 and Chars2 make sure that nothing is missed in an examination, thus improving the objectivity and replicability of forensic handwriting examination.

Accuracy: Chars1 and Chars 2 are 100% accurate at listing the characters in a text and characters overlapping in compared texts. Research testing document examiners performing this task manually shows that they can easily miss letters, especially letter combinations that are not words (such as “on” by itself vs “on” in the morpheme “-tion”.

Current Languages: English, Spanish, Arabic, Chinese, Greek, German, Dutch, Italian and others

Research-in-Progress Languages: None

 

DIFFER: DIFFERences in Compared Texts

Task: What are the differences (additions, modifications and deletions) between Document1 and Document2, 3, 4..?

Uses: DIFFER assists attorneys, law enforcement and human resource professionals.

Speed: DIFFER runs very quickly, in minutes.

Notes: In contract review or witness statement review, a primary task is finding out how much difference is there in versions.

Accuracy: DIFFER is 98% accurate at listing the differences in compared texts; the 2% error rate occurs when the documents are extremely different.

Current Languages: English, Spanish, Arabic, Chinese, Greek, German, Dutch, Italian and others

Research-in-Progress Languages: None