And; 53.5 for the DB3. However, in the DB4 the value is just 10.0 , probably because the donors are not used to signing in native Western styles. Signatures with either only a text name or only a flourish are found in proportions 5.1 and 8.3 respectively for the DB1; 33.3 and 16.7 for the DB2; 37.7 and 8.8 for the DB3 and; 90.0 and 0.0 for the DB4.PLOS ONE | DOI:10.1371/journal.pone.0123254 April 10,9 /Modeling the Lexical Morphology of Western Handwritten SignaturesAccording to the learning process, people in general use their own criteria to decide the number of lines and words in their signatures. In this paper the word distribution presented in the signatures has been counted. We find that in signatures with text, the dataset DB1, 90 are written in one line whilst for the remaining 10 it is two. For the 10 , the second line is often below the first and is usually shorter than first line and starts at about 20 of the signature length toward the right side. The rest of datasets contain all signatures drawn in one line. For those signatures written in one line, the proportion with one, two or three words is 50.0 , 36.0 and 14.0 respectively for the dataset DB1; 64.7 , 27.5 and 7.8 for DB2; 77.7 and 22.3 for DB3; and 92.5 , 7.5 for DB4. These two latter datasets do not contain signatures with three words. We are aware that the total number of words and letters per word depend on personal choice, influenced by the original name and surname. However, we note the influence of two surnames in the Spanish culture, since some signers include both. For signatures depicted in two lines, which applies only to BD1, the proportion is as follows: ?In the first line one, two and three words Quisinostat site appear respectively at 55.0 , 37.0 and 8.0 . ?In the second line, 71.0 , 21.0 and 8.0 of the signatures have respectively one, two and three words. For the number of letters per line, two types have been differentiated: type one is PX-478 structure related to signatures with one line (line 1) and type two related to signatures with two lines (line 2-1 refers to first line of a signature with two lines and line 2-2 to the second line). Type one is found in all datasets, whereas type two is a feature only of DB1. The parametric and actual number of letter distributions are displayed at Fig 4. For the type one, the KS test suggests that DB1 and DB2 are similar among themselves. This is also true of DB3 and DB4. Such a difference wasFig 4. Modeling the total letters per line PDF. doi:10.1371/journal.pone.0123254.gPLOS ONE | DOI:10.1371/journal.pone.0123254 April 10,10 /Modeling the Lexical Morphology of Western Handwritten SignaturesTable 2. Relationship between the text and the flourishes in the complex structures for the Western databases. Cases T + Fs + Fm Fs + T + Fm LFs + T + Fm FsLFs + T + Fm doi:10.1371/journal.pone.0123254.t002 DB1 7 22 68 3 DB2 0 27 37 36 DB3 5 37 47 11 DB4 0 0 0 0Table 3. Analytical results from Generalized Extreme Value distributions. Shape Letters line 1. DB1-DB2 Letters line 1. DB3-DB4 Letters line 2-1. DB1 Letters line 2-2. DB1 Slant All DBs Skew All DBs Corners (Fm) DB1-DB3 Corners (Fm) DB4 Corners (Fs) DB1 Corners (Fs) DB2 Corners (Fs) DB3 Text width All DBs Flo. width DB1-DB3 Width ratio DB1-DB3 Center ratio DB1-DB3 -0.30 -0.21 0.13 -0.34 -0.31 -0.09 -0.08 -0.09 0.50 -0.06 -0.33 -0.01 0.01 -0.02 -0.17 Scale 2.47 1.69 1.53 2.15 14.20 7.78 1.54 0.81 1.20 0.71 0.55 8.84 10.13 0.27 0.14 Local. 5.22 4.72 5.77 4.09 75.And; 53.5 for the DB3. However, in the DB4 the value is just 10.0 , probably because the donors are not used to signing in native Western styles. Signatures with either only a text name or only a flourish are found in proportions 5.1 and 8.3 respectively for the DB1; 33.3 and 16.7 for the DB2; 37.7 and 8.8 for the DB3 and; 90.0 and 0.0 for the DB4.PLOS ONE | DOI:10.1371/journal.pone.0123254 April 10,9 /Modeling the Lexical Morphology of Western Handwritten SignaturesAccording to the learning process, people in general use their own criteria to decide the number of lines and words in their signatures. In this paper the word distribution presented in the signatures has been counted. We find that in signatures with text, the dataset DB1, 90 are written in one line whilst for the remaining 10 it is two. For the 10 , the second line is often below the first and is usually shorter than first line and starts at about 20 of the signature length toward the right side. The rest of datasets contain all signatures drawn in one line. For those signatures written in one line, the proportion with one, two or three words is 50.0 , 36.0 and 14.0 respectively for the dataset DB1; 64.7 , 27.5 and 7.8 for DB2; 77.7 and 22.3 for DB3; and 92.5 , 7.5 for DB4. These two latter datasets do not contain signatures with three words. We are aware that the total number of words and letters per word depend on personal choice, influenced by the original name and surname. However, we note the influence of two surnames in the Spanish culture, since some signers include both. For signatures depicted in two lines, which applies only to BD1, the proportion is as follows: ?In the first line one, two and three words appear respectively at 55.0 , 37.0 and 8.0 . ?In the second line, 71.0 , 21.0 and 8.0 of the signatures have respectively one, two and three words. For the number of letters per line, two types have been differentiated: type one is related to signatures with one line (line 1) and type two related to signatures with two lines (line 2-1 refers to first line of a signature with two lines and line 2-2 to the second line). Type one is found in all datasets, whereas type two is a feature only of DB1. The parametric and actual number of letter distributions are displayed at Fig 4. For the type one, the KS test suggests that DB1 and DB2 are similar among themselves. This is also true of DB3 and DB4. Such a difference wasFig 4. Modeling the total letters per line PDF. doi:10.1371/journal.pone.0123254.gPLOS ONE | DOI:10.1371/journal.pone.0123254 April 10,10 /Modeling the Lexical Morphology of Western Handwritten SignaturesTable 2. Relationship between the text and the flourishes in the complex structures for the Western databases. Cases T + Fs + Fm Fs + T + Fm LFs + T + Fm FsLFs + T + Fm doi:10.1371/journal.pone.0123254.t002 DB1 7 22 68 3 DB2 0 27 37 36 DB3 5 37 47 11 DB4 0 0 0 0Table 3. Analytical results from Generalized Extreme Value distributions. Shape Letters line 1. DB1-DB2 Letters line 1. DB3-DB4 Letters line 2-1. DB1 Letters line 2-2. DB1 Slant All DBs Skew All DBs Corners (Fm) DB1-DB3 Corners (Fm) DB4 Corners (Fs) DB1 Corners (Fs) DB2 Corners (Fs) DB3 Text width All DBs Flo. width DB1-DB3 Width ratio DB1-DB3 Center ratio DB1-DB3 -0.30 -0.21 0.13 -0.34 -0.31 -0.09 -0.08 -0.09 0.50 -0.06 -0.33 -0.01 0.01 -0.02 -0.17 Scale 2.47 1.69 1.53 2.15 14.20 7.78 1.54 0.81 1.20 0.71 0.55 8.84 10.13 0.27 0.14 Local. 5.22 4.72 5.77 4.09 75.
