FORENSIC SYSTEM FOR DETERMINING CONTRADICTIONS IN WITNESS TESTIMONY.

CRIMINALISTIC SYSTEM FOR DETERMINING CONTRADICTIONS IN WITNESS TESTIMONY.

RATKIN Leonid Sergeevich, Candidate of Technical Sciences

In forensic science, a situation often arises when the testimonies of defendants, witnesses and other persons are confusing, contradictory and difficult to reconcile with other testimonies.

Currently, special automated systems can be used to determine contradictions and “inconsistencies”, the essence of which is as follows.

First of all, each testimony is drawn up in the form of a corresponding text file (TF).

For example, handwritten text is scanned and digitized, voice messages (audio files) are converted into text using special recognition systems, and photographs with evidence receive a verbal description.

Each TF in the information system (IS) has its own unique number (consisting of letters and numbers), allowing it (TF) to be uniquely identified in the IS among other files.

The resulting TF database allows, based on the unique number of each TF, to also uniquely identify any text fragment of an arbitrary file.

For example, the following structure can be used as a text block file identifier (TBFI):

• • TF identification number (size – from 5 to 10 bytes);
  • paragraph number in TF (2 bytes allow addressing 65,536 paragraphs in TF);
  • word number in paragraph (2 bytes address 65,536 words in each paragraph);
  • character number in word (1 byte addresses 256 characters in a word – letters and numbers).

If necessary, the sizes of the parts that make up the ITBF can be increased or decreased.

It is possible to take into account empty paragraphs (consisting only of the “carriage return” code) or ignore them (skip) when calculating the ITBF.

Example 1. 145C7 means the entire text file from the first to the last character.

Example 2. 5Ц83Р 00019 is the address of paragraph 19 of TF 5Ц83Р (all words and letters in the paragraph). Note that the method for constructing the ITBF does not assume numbering sentences (a sequence of words ending with a period).

Example 3. 71В42 00103 00052 is the address of word 52 in paragraph 103 of TF 71В42 (all letters and symbols in a word, “a word is considered to be a sequence of symbols limited by spaces on both sides”).

Example 4. 3С5Р7 00012 00007 00004 is the address of character 4, word 7, paragraph 12 of TF 3С5Р7. In particular, for the word “2005, character number 00004 means the number “5”.

ITBF can specify both the beginning and the end of a text block in a file. It is possible to specify the addresses of the beginning and end of a block in one ITBF separated by a dash, in which case the text block (i.e. the entire text) between the first and second ITBFs is implied.

Example 5. 23U34 00023 – 00070 means a text fragment from paragraphs 23 to 70 of TF 23U34.

Example 6. A45P5 00101 00075 – 00099 – corresponds to a text fragment from word 75 to 99, paragraph 101 of TF A45P5.

Example 7. C12T2 00012 00034 00001 – 00039 00075 00009 denotes the address of a text fragment from the 1st letter of the 34th word of the 12th paragraph to the 9th letter of the 75th word of the 39th paragraph of TF C12T2.

The order of the constituent parts of the ITBF can be changed, for example, first the paragraph number in the TF, then the word number in the paragraph, then the letter (or character) number in the word, and finally the TF identification number.

Since the ITBF consists of 4 parts, any of the 24 combinations (4! = 24) can be used to form an identifier.

More complex constructions that include ITBF elements can also be used.

Note that the calculation of the ITBF for an arbitrary text block (TB) can be performed automatically by a computer program for working with text, by selecting the TB with a mouse-type manipulator during viewing.

Thus, with the help of the ITBF, each TB has a unique address that uniquely identifies the TB among many fragments of other files. The developed mechanism for addressing TBs can be used to construct a matrix of the degrees of consistency of TB files.

Let there be Nreadings (continuous fragments of TF), the mutual inconsistency of which must be established. Let ai denote the corresponding ITBFs, i = 1…N. Let bij be the degree of mutual consistency of two TBs for ITBFs ai and aj, which takes the value 1 if two TBs are mutually consistent, and 0 if a contradiction exists (is established), the decision on the value of the degree of consistency is made by expert specialists (for example, within the framework of a collegial body). The following construction of the TB consistency degree matrix is ​​possible:

d> 

…

Since the readings are assumed to be internally consistent, the diagonal of the consistency matrix (highlighted in gray in the figure) contains ones.

It is possible to set another value that denotes the internal consistency of the readings, for example 100 – it is important that the value of the degree of internal consistency is the same for each of the readings.

The property of the constructed matrix is ​​its symmetry with respect to its main diagonal, i.e. the transposed matrix is ​​equal to the original one, since determining the degree of consistency of two TBs is commutative.

It is possible to determine the total degree of external consistency of each TB (by rows or by columns – it does not matter, the result will be the same) with respect to the other blocks: Bi = bi1 + bi2 +…+ biNor Bi = b1i + b2i +…+ bNi, i=1…N, i=1…N.

The Bi characteristic can be called “the relative consistency of the current, i-th TB in relation to other TBs. When calculating Bi for i = 1…N, it is possible to determine the largest and smallest values ​​of Bmax and Bmin, as well as order Bi, i = 1…N, in ascending or descending order.

Since each degree of consistency (see matrix) takes the value 1 if there is no contradiction, and 0 if there is a contradiction (according to experts), then the maximum value Bmax corresponds to the most consistent with the other fragments of the TB readings, and the minimum Bmin – to the least consistent with the other (most incorrect) reading.

The resulting ordered sequence Bi, i = 1…N, in ascending order of Bmin …Bmax, in this case represents a sequence of clarification of incorrect readings from the persons who presented them.

In automated forensic systems, it is proposed to use modern systems for recognizing logically contradictory constructions, self-learning systems, systems for automatically searching for contradictions in testimony, and other developments.

A highly intelligent set of systems for searching for internal and external contradictions in testimony will allow experts to automatically make decisions on the degree of contradiction in the corresponding text blocks of files.

The scale of the task requires intensive use of distributed computing technology based on production clusters [1].

Conclusions

1. It is possible to specify the degree of consistency on a continuous numerical interval [c, d] and determining the weighted average of expert assessments. It is also advisable to use data reliability coefficients that allow determining the degree of plausibility (greater confidence) in the testimony of certain witnesses and other persons.
2. The TB consistency degree matrix is ​​also suitable for examining internal contradictions in testimony. In this case, the diagonal of the matrix contains not units, but the corresponding expert values, which generally does not affect the operation of the mechanism for determining the most incorrect testimony.
3. The contradiction determination system can be used in other areas, for example, to assess shortcomings in regulatory documents in various fields, to analyze incompatible technical and technological solutions in industrial production, etc. [2].

References

— Ratkin L.S. Method for constructing a subject area basis based on its basic definitions. /Problems of Defense Technology, No. 1 (320), 2004

— Ratkin L.S. Principles of Determining the Degrees of Internal and External Consistency of Parts of Regulatory Documents. /Industrial Policy in the Russian Federation, No. 5, 2005.