=Paper=
{{Paper
|id=Vol-1975/paper21
|storemode=property
|title=Adaptation Algorithm of the Computer Text Font Size for Optimal Perception
|pdfUrl=https://ceur-ws.org/Vol-1975/paper21.pdf
|volume=Vol-1975
|authors=Alexander Varnavsky
|dblpUrl=https://dblp.org/rec/conf/aist/Varnavsky17
}}
==Adaptation Algorithm of the Computer Text Font Size for Optimal Perception==
https://ceur-ws.org/Vol-1975/paper21.pdf
Adaptation Algorithm of the Computer Text
Font Size for Optimal Perception
Alexander Varnavsky
Ryazan State Radio Engineering University, Ryazan, Russia,
varnavsky alex@rambler.ru,
http://www.rsreu.ru
Abstract. The paper presents the result of the research of the influence
of text font size on attention indicators. On the basis of the experimental
data, the multiple linear regression of the dependence of the optimum of
the font size on the criterion of maximizing the value of mental efficiency
indicator from the indicators of attention and memory of the subject was
constructed. An algorithm for adapting the font size of text for optimal
perception is presented.
Keywords: memory, attention, font size, mental performance, psychophys-
iological testing, optimal font size, forecasting model, adapting algo-
rithm.
1 Introduction
Perception is an important part of the communicative process. It affects the effi-
ciency of receiving and assimilation of information flows. Sometimes perception
is connected with examining of certain objects, which are basically combina-
tion of imaginary and text information. Increase of information volumes and
appearance of new types of devices results in searching of effective items of its
representation [1].
It is important to consider the influence of parameters of the provided text
information, in particular a type of a font and its size, on perception of the text.
However, there is no information on personification of type size in relation to
the specific person.
Perception of text information and level of its assimilation depends on mea-
sure values of cognitive processes of the person working with it. That is why
the current values of such indices can be selected as criterion of change of type
size. There is a sense to stop on a choice only of processes of memory and atten-
tion as they generally cause perception of information and its storage, which is
important in process of work with information.
The aim of the work is the research of influence of type size of the provided
information on indices of attention and creation of the algorithm for adapting
the font size of text for optimal perception.
To achieve this goal, the experimental research will be carried out. The stages
of the research will be as follows.
1. The choice of psycho-physiological tests for the evaluation of indicators of
cognitive processes. Use the test containing letters of different font sizes.
2. Carrying out an experiment and testing a group of subjects.
3. Data collection and processing.
4. Building the regression model of dependence of optimum type size on indexes
of cognitive processes of the examinee.
5. Development of the adaptation algorithm of the text font size for optimal
perception.
We can consider the works similar to this theme.
In [2], Peng Chengyuan et al. present an adaptive text extraction method
to represent text on devices with a small display which can automatically ex-
tract key information from original text and keep semantic meanings as close
as possible. The authors suggest combine both statistical methods and coarse
coding algorithm from neural science to shorten long text sentences in terms of
generalization. In [3], Benedicenti Luigi et al. present an expert system that can
support dynamic interface adaptation. This will allow an ubiquitous applications
to reformat their interface depending on the screen real estate they are granted.
Basically human examining of different objects connected with certain reg-
ularities of perception and attention and can be subjected to errors and distor-
tions [4], [5], [6]. Some of them can lead to different negative phenomena and
consequences. Adverse conditions for perception of information flows increase an
operating time and reduce the volume of the acquired information. That is why
researching of features of information perception including text information and
creating optimum working conditions with information and its representation is
an actual aim.
Perception of the examining information depends on its representation. For
example, in [7], [8], Andreeva O.N. describe influence of location of information
screen objects of the monitor on time and accuracy of their recognition. This and
other features should be considered in case of design of new telecommunication
interfaces [9]. In [10], Koumpis A. et al. offer the use of the text media type and
present possible adaptations of it, with respect to the particular needs, abilities
and preferences of diverse user categories, including disabled people. Moreover
in case of design and using of electronic items it is important to consider specific
and psychophysiological features [11], [12], attention index [13], [14].
Due to the development of electronic studding systems many labors are de-
voted to increasing of efficiency of submission of educational information [15].
It is one of the directions of the concept called by pedagogical design [16], [17].
It is used for creating study curse, its design and interface, creating educational
materials.
The research of use of special fonts in systems of visual navigation that help
to be guided in the best way in city space is described [18]. The legibility of
fonts for design of texts of official papers by means of reading time dimension is
research [19]. In this result the regression model of computation of a rank of a
font in its geometrical parameters is described [19].
Research of features of perception of the most often used type fonts [20]
based on methods of electroencephalography and oculography shows that type
fonts differ not only graphic signs, but also have the psychophysiological specifics
defining quality of perception of the text, his understanding and storage. There
is the conclusion concerning what fonts it is the most rational to use by criteria
of the largest speed of reading, readability, smaller exhaustion of the visual
analyzing and the spent cognitive efforts is drawn.
On the basis of an assessment of legibility of fonts in electronic issuing by
method of paired comparing and by method of speed sensing of reading in [21]
revealed that grotesque fonts have advantage over serif types.
The tachistoscopic experimental research of visual sensation of meaningless
text information on the display are carried out [4]. Dependence of perception of
the information on type size which is characterized by certain preferable strategy
of relocation of eyes is shown.
2 Choosing of psychophysiological tests
Indices of attention and short-term memory can be determined by results of
psychophysiological testing with use, for example, of proof test [22] and the test
for storage of numbers [23]. The main reason of a choice of proof test is that
the training practically does not exert the considerable impact on test results.
Besides, the possibility of further implementation and automation of conducting
testing in an information display system was considered.
In case of a choice of option of proof test stopped on Burdona-Anfimov’s
test as it can be easily automated (in comparison with Landolt’s rings and tests
with Ivanov-Smolensk modification) and its execution is a little more difficult
for the adult user, than at Burdon’s test [22]. It is important for an assessment
of abilities of examinees of different level for a short period of time.
Correction test of Burdon-Anfimovs [22] is the technique intended for a re-
search of stability and productivity of attention, calculation of an index of fa-
tigue, determination of mental working capacity. This technique consists in pre-
sentation to the examinee of lines from a random series of letters of the Russian
alphabet which it shall view and eliminate two given letters sequentially. Results
of test are estimated by quantity of the passed signs, on runtime or by quantity
of the viewed signs [22].
The results of Burdon-Anfimov’s test define main indexes of attention: in-
dex of mental working capacity Amp , attention performance index Ap , level of
concentration of attention Aca , quality of working Aqw [22]:
N M − (O + P )
Amp = · , (1)
t n
N
Ap = , (2)
t
S−P −O
Aca = , (3)
n
O+P
Aqw = 1 − , (4)
N
where N is number of characters in the part of proof test worked by the exam-
inee; t is runtime of the test in seconds; M is total number of the crossed-out
characters; O is number of incorrect crossed out characters; S is number of cor-
rectly crossed out characters; P is number of the passed characters; n is number
of characters which needed to be eliminated in the viewed part of proof test.
These indexes characterize operation with text information. So the perfor-
mance measure of attention corresponds to number of the letters viewed in unit
of time, an index of mental working capacity is to number of the letters viewed in
unit of time taking into account existence of errors. The figure of merit of opera-
tion is equal 1 in the absence of errors and decreases in case of their appearance
of subjects more, than the number of the passed and incorrect eliminated charac-
ters are more. Level of concentration of attention shows a share of truly crossed
out letters from total number of characters which needed to be eliminated.
For testing of abilities of a short-term memory is selected the test for storage
of numbers. This test consists in presentation to the examinee of 10 different
double-valued numbers which it shall reproduce on memory after their viewing
within 30 seconds [23]. Coefficient of a short-term memory Ksm it is calculated
by a formula:
K
Ksm = , (5)
L
where K is the quantity of correctly reproduced double-valued numbers, L is
number of the initial numbers.
3 Procedure of the research
For the research of influence of type size of the text on indices of attention and
creation of a mathematical model of dependence of optimum type size on indices
of attention and memory of the examinee the following series of experiments was
conducted:
1. Carrying out method
(a) Participants. Participants of an experiment were 30 students of 3-5 courses
of the Ryazan State Radio Engineering University. Number of males is
15, women’s is 15. Average age of participants is about 21 ± 0.8 year.
(b) Materials. For carrying out an experiment 4 types of forms of Burdona-
Anfimovs differing only in type size of letters were used 10 pt., 12 pt.,
14 pt., 16 pt. Also was used the record of 10 double-valued numbers
(L=10).
(c) Procedure of the research. Experiments were made in the first half of
day in 3 days at groups till 8-12 of people. External conditions at all
groups were created identical. After briefing and explanations of a sense
of an experiment examinees passed trial test for memory and attention.
At the same time examinees were told two letters which they shall look
for and eliminate in attention tests. Further record of 10 double-valued
numbers was shown and the memory test was executed. Then forms of
the test of Burdona-Anfimov which type size 10 pt. were distributed and
examinees were doing it within 5 minutes. Tests with type size 14 pt.
and 16pt. were similarly executed.
2. The results
For each examinee by results of execution of proof test and the test for
memory the set of the following values used in formulas (1)-(5) was created:
N , M , O, S, P , n, K. The results are in the Table 1.
Table 1. Results of the experiment (Average value)
Index Average value For font size 10 For font size 12 For font size 14 For font size 16
N 459.7 ± 35.6 449.5 ± 34.6 444.9 ± 48.3 465.9 ± 23.0 478.4 ± 20.4
M 26.4 ± 6.7 29.0 ± 9.1 24.3 ± 5.6 28.1 ± 4.2 24.1 ± 5.8
O 0.03 ± 0.2 0.03 ± 0.2 0.03 ± 0.2 0.03 ± 0.2 0.0 ± 0.0
S 26.7 ± 6.5 30.1 ± 7.8 24.3 ± 5.6 28.2 ± 4.2 24.1 ± 5.8
P 3.0 ± 2.6 3.3 ± 3.0 2.7 ± 1.9 3.3 ± 3.2 2.7 ± 2.0
n 29.7 ± 6.0 33.5 ± 6.8 27.0 ± 5.2 31.5 ± 2.8 26.7 ± 5.8
K 6.7 ± 0.24
3. Processing of results of the experiment
Processing of results of the experiment was carried out in a statistical packet
of R. By results of the experiment on the basis of values N , M , O, S,
P , n, K for each examinee, is created the set of 30 observations with the
following variables calculated by formulas (1)-(5): Aca1 , Amp1 , Ap1 , Aqw1
where x = 10;
Aca2 , Amp2 , Ap2 , Aqw2 where x = 12;
Aca3 , Amp3 , Ap3 , Aqw3 where x = 14;
Aca4 , Amp4 , Ap4 , Aqw4 where x = 16;
Ksm ;
where x is type size in Burdona-Anfimov’s test.
In the received results the dispersion of values was watched. We will mark
that the task is directed to creation of model for prediction of optimum type
size on indices of memory and attention. Therefore the measure values received,
for example, after classes or in the evening when in most cases measure values
of cognitive processes are lowered, can be also used for prediction of optimum
type size. We visualize the received attention measure values with the boxplot
(Fig. 1). This figure shows the median (bold line in rectangles), quantiles and
the values located between them (rectangles) and range (line with the mustache)
of the corresponding indicators and illustrate the results obtained as a result of
the experiment.
Histograms of attention indexes based on proof test results with the type 10
pt. are represented on the Fig. 2. Due to the Shapiro-Wilk test value distributions
Fig. 1. The boxplot of the received attention index in case of different type sizes
of indexes of mental working capacity and productivity of attention are normal
(p > 0.05). Value distributions of indexes of attention concentration level and
accuracy of operation aren’t normal and are offset to the right. It is connected
to the fact that the greatest possible value of these indexes is equal to unit and
corresponds to faultless execution of proof test, and the most part of examinees
executed this test correct or almost correct. It isn’t excluded that if time of
carrying out an experiment was big, then it would lead to increase in number of
errors owing to rise of exhaustion and as result, to offset of value distribution of
these indices to the left and its approaches to normal.
4 Regression model of dependence of optimum type size
on indexes of cognitive processes of the examinee
In general it is possible to set different optimization tasks on operation with
text information, for example, minimum time of viewing, the minimum number
of errors, the maximum volume of viewing.
The minimum values of such parameters of operation will lead to maximizing
values of certain indices of attention.
Therefore in case of creation of regression model it is necessary for each
examinee to define type sizes y1 , y2 , y3 , y4 in case of which the maximum values
of the appropriate indices of attention are watched:
Fig. 2. Histogram of distribution of indices of attention Aca1 , Amp1 , Ap1 , Aqw1
at y1 obtained max(Amp1 ; Amp2 ; Am p3; Amp4 );
at y2 obtained max(Ap1 ; Ap2 ; Ap3 ; Ap4 );
at y3 obtained max(Aca1 ; Aca2 ; Aca3 ; Aca4 ); (6)
at y4 obtained max(Aqw1 ; Aqw2 ; Aqw3 ; Aqw4 ).
Let us set the task of a prediction of optimum type size y in case of which
minimum time of viewing of the text with the minimum number of errors is
watched, that corresponds to the maximum index of mental working capacity,
on measure values of proof test of a certain type size, for example 10 pt.
yopt = f (Amp1 , Ap1 , Aca1 , Aqw1 , Ksm ). (7)
So, the set of values of the predicted variable will consist of sets of y1 values of
each examinee. Set y1 was chosen with the figure that it based on the Amp1 value
that calculates both the operating time, and number of the made mistakes.
Values Aca1 , Amp1 , Ap1 , Aqw1 are defined from values N , M , O, S, P , n, K,
which were received in tests. From the point of view of formulas (1) and (2) the
value Ap1 is a part of Amp1 . To estimate a possibility of switching on of these,
as well as others, indices at model, we will realize verification of presence of
multicollinearity in the coefficients of inflation of dispersions this by determina-
tion between couples of values. We have the maximum value of such coefficient
between Aca1 and Amp1 equal 1.31. Therefore, the problem of multicollinearity
is absent and it is possible to include all considered attention indices in model
Except checking out on multicollinearity, there was the check of premises of
the linear model.
As a result of regression analysis multiple linear regression of a look was
received
yopt = a0 + a1 · Aca1 + a2 · Amp1 + a3 · Ap1 + a4 · Aqw 1 + a5 · Asm (8)
with R2 = 0.4329 (F (5, 24) = 3.664, p < 0.05).
More successful model from the point of view of R2 value, which is differ
from value in model (8) is multiple linear regression with interactions:
yo pt = a0 + a1 · Aca1 + a2 · Amp1 + a3 · Ap1 + a4 · Aqw1 + a5 · Ksm +
+a6 · Aca1 · Amp1 + a7 · Amp1 · Aqw1 + a8 · Ap1 · Aqw1 (9)
with R2 = 0.6418 (F (8, 21) = 4.703, p < 0.01).
In that model coefficients have values:
a0 = −460.86, a1 = −7.65, a2 = 117.34, a3 = 177.89, a4 = 479.5, a5 =
−0.58, a6 = 7.39, a7 = −112.19, a8 = −179.66.
So, the model which allows to predict optimum type size of the text for
obtaining the maximum level of mental working capacity on indices of memory
and attention, defined in case of type size 10pt. is received. In case of obtaining
the fractional values of type size received on model it is necessary to carry out
its rounding to the next value to within 0.5.
5 Adaptation algorithm of the text font size for optimal
perception
The received model of dependence of optimum type size on indices of cogni-
tive processes of the examinee can be used in perspective display systems of
information which have potential of personification and can change automati-
cally parameters of the displayed information depending on personal properties
and the user’s indices. In this case such systems shall possess modules for an
assessment of measure values of memory and attention.
The following algorithm of operation of a perspective display system of in-
formation with the built-in software module exercising control of type size of the
provided text information is offered:
1. After user login it is offered to pass Burdona-Anfimov’s test with type size
10pt., and then the test for storage.
2. By results of passing of these tests on formulas (1)-(5) there is a calculation
of measure values Aca1 , Amp1 , Ap1 , Aqw1 , Ksm .
3. Proceeding from the received measure values of attention and memory, on a
formula (9) is performed the calculation of best value of type size ypt .
4. The output of text information is carried out with type size ypt .
5. If works is carried out with the text for a long time and probably rise of
exhaustion, then the user is offered to pass repeatedly these tests for the
purpose of obtaining new value ypt . The duration of a period of the continu-
ous working with information, on which termination is necessary to multiple
definition of values Aca1 , Amp1 , Ap1 , Aqw1 , Ksm ypt depends on individual
users features. The assessment of its value is a subject of future research.
In the offered algorithm the model is used (9). It was received with test of
Burdona-Anfimov on paper forms. According to researches reading the text from
the screen is carried out for 25% more slowly, than at paper form. It will be visible
in attention indexes Amp1 and Ap1 . The model will predict best value of type
size in case of such values. And if the monitor reproduces information 1:1, then
the received best value can be used without any changes. In general it is planned
to conduct the research specifying this moment and need of adjustment of the
predicted value of type size depending on monitor type.
6 Conclusion
Efficiency of perception of computer information is influenced by a set of factors,
including its representation and users psychophysiological status.
The research conducted in labor on the example of use of proof test showed
change of measure values of attention in case of change of type size of the shown
text. On the basis of the data obtained as a result of an experiment the model
of dependence of an index of mental operability of type size, optimum from the
point of view of maximizing, on indexes of attention and memory is constructed.
For increase in coefficient of determination it is necessary to include in model
bigger quantity of factors, including measure values of other cognitive processes
and parameters of text information. Also the forecasting accuracy of optimum
type size can be raised due to specification of coefficients of model when carrying
out bigger number of experiments, including using bigger number of the sizes of
fonts.
The received model allows predicting optimum type size for the specific user
and can be used in perspective display systems of information for personifi-
cation of parameters of the output text information. As an example of such
systems where the problem of optimization of layout of information and appli-
cation of the offered algorithm of adaptation of type size is urgent, it is possible
to give systems of e-learning and interactive electronic technical manuals with
which different experts work at post-production stages of life cycle of difficult
knowledge-intensive products.
Further working in this direction assumes carrying out experiments taking
into account bigger number of factors. For example colors, line spacing, the
size of area of the screen on which the text is provided etc. Also there will be
research of influence of exhaustion on value of optimum type size. Some of such
researches are planned to be conducted with use i-tracker, realizing the analysis
of movement of a look of the examinee. The statement of criterion of optimization
of information representation not only on the basis of maximizing one index of
attention, and on the basis of integral criterion is of interest.
Moreover, more difficult process can be considered. Such as understanding,
including not only the perception of information, but also its comprehension and
the analysis.
References
1. Belko, T.V., Sasin, D.G.: Font and image in the history of visual communications
Izvestiya of the Samara Scientific Center of the Russian Academy of Sciences. 2010.
Vol. 3–1. Pp. 236–241.
2. Peng, Chengyuan, Vuorimaa, P.: Text Adaptation for Mobile Digital Teletext Web
Intelligence, 2004. WI 2004. Proceedings. IEEE/WIC/ACM International Confer-
ence on. 2004. Pp. 453–456.
3. Benedicenti, Luigi, Petty, Sheila: Text Interface Adaptation in Ubiquitous Appli-
cations 2012 Sixth International Conference on Innovative Mobile and Internet
Services in Ubiquitous Computing. 2012. Pp. 311–316.
4. Beloscova, K.V., Artemenkov, S.L.: Experimental study of the order of perception
of textual information on the display screen. Experimental Psychology in Russia:
Traditions and Perspectives. 2010. Pp. 230–234.
5. Droit-Volet, S., Lamotte, M., Izaute, M.: The conscious awareness of time distor-
tions regulates the effect of emotion on the perception of time Consciousness and
Cognition. 2015. Vol. 38. Pp. 155–164.
6. Pan, Yi, Hou, Xiu: Working memory distortions of duration perception are mod-
ulated by attentional tags Acta Psychologica. 2016. Vol. 165. Pp. 48-52.
7. Andreeva, O.N.: Reducing the time of recognition of the information on the monitor
screen to the student. In the world of scientific discoveries. 2011. Vol. 9.1 (21). Pp.
267–277.
8. Andreeva, O.N.: Accuracy of recognizability of the test image in different points
of the field of view of the operator. Bulletin of Neurology, Psychiatry and Neuro-
surgery. 2011. Vol. 8. Pp. 57–63.
9. Artemenkov, S.L.: Psychology of perception and development of new telecommu-
nication interfaces. Telemultimedia. 2004. Vol. 4 (26). Pp. 15–19.
10. Koumpis, A., Karagiannidis, C., Stephanidis, C.: Adaptations of the text media
type: addressing the user’s perspective Proceedings International Workshop on
Multimedia Software Development. 1996. Pp. 160–168.
11. Andreev, V.N.: Psychological aspects of the presentation of information on the
display screen in automated learning systems: the author’s abstract of the thesis
for the degree of candidate of psychological sciences. St. Petersburg. 1991. 16 p.
12. Oparina, N.M.: Influence of psychophysiological characteristics of trainees on the
effectiveness of their work with the use of ASO Pedagogical Informatics. 2004. Vol.
2. Pp. 81–88.
13. Koneva, L.V., Plotnikov, V.V.: Properties of attention as a function of controlling
human behavior. Vestnik of new medical technologies. 2009. Vol. 3. Pp. 137–139.
14. Novikova, N.M., Aleksandrov, A.S.: Automation of research of the human opera-
tor’s attention properties in the human-display system. Bulletin of VSTU. 2012.
Vol. 7–1. Pp. 43–45.
15. Gura, V.V.: Philosophical aspects of the pedagogical design of the electronic media
education environment. Philosophical Problems of Information Technologies and
Cyberspace. 2011. Vol. 2. Pp. 84–96.
16. Nagaeva, N.A.: Pedagogical design and pedagogical design: problems and
prospects. Informatization and communication. 2012. Vol. 4. Pp. 61–64.
17. Vuvarov, A.Yu.: Pedagogical design. Computer science. 2003. Vol. 30. Pp. 2–31.
18. Bekishev, M.A.: Font in visual navigation systems Society. Wednesday. Develop-
ment (Terra Humana). 2011. Vol. 4. Pp. 123–127.
19. Tokar, O.V., Zilbergleit, M.A. Litunov, S.N.: Evaluation of the legibility of the font
on the material of the official document. Omsk scientific herald. 2009. Vol. 2–80.
Pp. 246–249.
20. Morozova, L.V., Murin, I.N.: Psychophysiological specificity of perception of
printed type. Bulletin of the Northern (Arctic) Federal University. Series: Natural
Sciences. 2013. Vol. 3. Pp. 76–85.
21. Vasyuta, S.P., Hamula, O.G.: Influence of speed of reading of a font on convenience
of perception of the text in electronic editions. Internet-journal Naukovedenie.
2013. Vol. 3 (16). P. 56.
22. Brunner, E.Yu.: Better than superweaving. Methods of diagnosis and psychocor-
rection. Rostov-on-Don: ”Fenix”, 2006. 316 p.
23. Karelin, A.A.: Large encyclopedia of psychological tests. Publishing house:
”Exmo”, 2007. 416 p.