How can we provide designers and developers with some support to identify the most appropriate gestures for gestural user interfaces depending on their context of use? To address this research question, GEStory was developed to address this research questions. It is an on-line atlas of gestures resulting from gesture elicitation studies with four main functionalities: (1) search for user-defined gestures identified in such studies by querying its features in an interactive design space, (2) show the preferred gestures and their characteristics for a given action (represented through a referent) with a given device in an environment and/or carried out with various body parts, (3) compare the existing studies and (4) suggest adding new studies. To feeGdEStory, two Systematic Literature Reviews (SLR) were performed: a macroscopic analysis of 216 papers on their metadata, such as authors, definitions, year of publication, type of publication, participants, referents, parts of the body (finger, hand, wrist, arm, head, leg, foot, and whole body), number of proposed gestures; a microscopic analysis of 267 papers analyzing and classifying the referents, the final gestures coming out the consensus set, their representation and characterization. It also proposed an assessment of credibility of these studies as a measure for categorizing their strength of impact.GEStory acts as an interactive design space for gestural interaction to inform researchers and practitioners on existing preferred gestures in diferent contexts of use, and enable the identification of gaps and opportunities for new studies.
santiago.villarreal@uclouvai(nS..bVeillarreal-Narvaez) overlapping, rarely incremental and complementary.
Let us consider an example showing how a GES is conducted: Suppose the case that a designer wants to develop a gesture user interface for a new smartwatch prototype that allows users to control 5 actions (materializerdefearsents) in the environment of a smart car: (t1) turn on the radio, (t2) turn on air conditioning, (t3) answer a call, (t4) turn of the radio, and (t5) turn of the air conditioning. For the study a group of potentia l uisebrrsought together, perhaps 20 in number,i.e.,| | = 20. Each participant is shown the context existing before and after performing each action:e.g.,The radio is of (before) and the radio is on for the referent t1 (after). The participant is asked to propose a gesture command using the smart watch to execute the referred action. At the end of the GES, the designer, who compiled a set of 100 gesture proposals = 5 (referent×s)20 (proposals), now looks at the set of gestures elicited for each function to understand which gestures are in agreement among participants. If the agreement is substantial and the sample of participants is representative enough of the target population that the gesture-based interface is targeting, then the designer is expecting that the prompts are intuitive and that other users, who were not part of the GES, are likely to guess, to learn easily, and hopefully to prefer the same types of gestures. Although these GES are num2e]r, otuhsey[ do not answer all design questions as they do not cover many contexts of use. Consequently, the following research question emerges from this paradoxical situation:
RQ=how can we provide designers and developers with some support to identify the most appropriate gestures for gesture-based user interfaces depending on their context of use?
The expected contributions of our doctoral thesis are as follows:
1. Two SLRs related to gestural interaction: one on GES metadata and one on gestures
characteristics, along with a classification and discussiono.nA-lnine Zotero collectioofn
relevant papers and classification.
2. New methodological aspects for conducting a GES and new G4,E5S, 6[
Regarding GES surveys and reviews, there are only three such studies: Veutlaelt.i[c10]
conducted a systematic literature review (SLR) on hand gestures for user interfaces, but without
having GES as the focus of their investigation; Vogiatzidakis and Koutsabasis first performed a
GES review [
Regarding software tools, there are several software pursuing diferent goals than ours.
GestMan 1[
Since our research question is not fully addressed, our initial idea consists of compiling and consolidating the results of all existing GES into a single gesture repository that can be queried to address the research question. This repository becomes a gesture atlas where each gesture is characterized according to several dimensions: user, task, device, environment, human limbs, etc. For this purpose, our research method is structured in five connected stage1s):(Fig. Existing GES =21-682p2ap9pa8eprpesarispneecrlxuscdleueldidgeibdle 1,819papersthrough
databasesearching +430papersthrough
othersources -311duplicatepapers =1,938papersidentified 1,938papersscreened -1,640irrelevantpapers =298papersafterscreening
=26-78pp2aa7pp5eerprssaipenexccrlusludedeleidgdible SLR of GES metadata
SLR of gesture characteristics anNdemweGthEoSds 1,394papersthrough
databasesearching +422papersthrough
othersources -301duplicatepapers =1,515papersidentified
GEStory -NeDwifcfeornetnetxtusseorfsuasnedtasks RDeoqmuairienmmeondtesl - Differentdevicesandsensors DDeevsieglnopment - Differentenvironments Newmethodologicalapproaches - Credibilityindex - Referentlesselicitation - Alternativemeasures - Modalitytransfer
Validation QueryingGEStoryby - Usertypeandcharacteristics - Tasktype - Device,sensortype - Environment - Humanlimb - Combination 1,515papersscreened -1=,224705irpraepleevrasnatftpearpsecrrseening
SLR of GES metadata:conduct a first SLR based on the metadata describing each G2E],S [ such as the year of publication, venue, number of participants, number of referents, experimental setup, number of proposed gestures and number of consensus gestures. Our approach was inspired by the four-phase SLR method (Identification, Screening, Eligibility and Inclusion) proposed by Liberateit al. in [19] and the flow was represented in a PRISMA diagram. For identification , the queryQ = (”Gesture” AND ”Elicitation” AND ”Study”)was performed on ifve major Computer Science digital librariei.se.(,ACM DL, IEEE Xplore, Elsevier ScienceDirect, Elsevier Ei Compendex, and SpringerLink) and other souric.ee.,sD(BLP CompleteSearch and Google Scholar) to identify 2,249 candidates, from which 311 duplicates were eliminated. For screening, we retained only those papers that explicitly introduced a GES for UI design, discussed a GES, or explicitly used a method to examine a GES, thereby leaving 298 paperesl.igFiobirlity, 82 papers were excluded that did not match 3 conditions, leaving a final co r=p2u1s6of studies for our examination. Foinrclusion, we verifyied quantitative and qualitative aspects of our corpus of papers, stored and maintain asoann-line collectiownithZotero,a multi-platform bibliography management.
SLR of gesture characteristi.cSsince the first SLR focused on GES metadata only, we ran a
second SLR to provide an in-depth analysis of gestures elicited and agreed upon in GES after
updating the collection, stopped at the beginning of 2021. We follow the same methodology
with theQ = (”Gesture” AND (guess* OR elicit*) AND (study OR experience):)1,816 papers
were firstly identified, 1,515 papers were screened after duplicates removed, 275 papers became
eligible, and 267 papers were finally included. Based on these SLRs, we have obtained concepts
and terminologies that are part of the GES studies, we discussed some examples of representative
GES, and we provided data and calculations such as the average, the mean, maximum, and
minimum of the number of participants, references, collected gestures, final gestures, etc.
Moreover, consensus gestures are classified according to several dimensions: a taxonomy of
referents based on task classificatio2n0][, a classification of 3D gestures21[], a classification of
gestures in Augmented Realit2y2][ and another one for the whole body to control a humanoid
robot 2[
New GES and methods. To complement our repository, our search conditions focus in
new contexts such as diferent users and tasks, diferent devices and sensors and diferent
environments, we identified some areas uncovered by existing GES and subsequently conducted
some of them, such as for head and shoulders gestur9e],sf[or zenithal gesture5s],[for
radarbased gestures6[
Development ofGEStory [27]. Based on the results of the two aforementioned SLRs, a domain model has been defined (Fig.2) to create the databaseGoEfStory, an on-line gesture atlas for querying GES on multiple criteGrEiaSt.ory is presented as an interactive design space, such as the one for wearable device2s8][, where various design dimensions can be explored. In particular, selecting any particular human limb should result in selecting GES satisfying this criterium (see a prototype in F3ig).. GEStory, beyond making gestures accessible, should also provide some guidelines in selecting and designing gestures based on its29a]talansd[, possibly, automate its evaluation based on guidel3in0e].s [
The framework used is Vue.js. The Vue file format is divided into 3 parts: 1. The HTML structure of the page or element that you want to display, 2. The JavaScript methods and state variables used by the component. This part also allows use the Listener design pattern to push changes to other components, 3. The CSS part that corresponds to the style of the component.
GEStory has a total of 3902 gestures from 267 gesture elicitation studies (GESs) obtained in the 2 SRLs, the information of these gestures are public and availabledianttah.jseo nfile
The architecture of the main page is divided into 3 parts (Se3e)Fwigh.ich communicate with each other. These 3 components all inherit from the Vue.component class: • The bodyMap component which takes care, based on data extracted frodmattah.jeson ifle, of displaying points representing gestures on the body map. It has diferent attributes such as the list of body regions (bodyRegions). As for its methods, it mainly has methods 1
BodyPart BodyPartId BodyPartName BodyPartDesc Location
Device DeviceID DeviceName DeviceDesc
User UserId Number SDAge MeansAge MFRatio
Is replication of, Is generalization of, Is repurposing of
Replication, Generalization, Repurposing RefID RefName RefRepresentation RefGeneric
1,n
Gesture GestureId GestureName GestureRepresentation GestureImage GestureType GestureForm GestureNature GestureSymmetry GestureLocale Agreement 1,n 0,n
GEStudy StudyId StudyTitle Year
Published 0,n AUuRtLhors
StrengthOfEvidence Replication Generalization Repurposing 0,n 1,n 1,n 1,n 1
FunctionSubType TaskID TaskName TaskDesc
1,n
FunctionType TypeID TaskSetName TaskSetDesc
Environment
EnvDesc Process
PreTest 0,n TPeosstTest
Consolidation
related to the display of points on the body map (getPositionForTypeAndItem, drawLineR, changeBodySelection for example). • The DataFilter component takes care of displaying filters as well as displaying the list of gestures extracted from the sadmaeta.json file. Each gesture in the list has the name of the gesture, the name of the study andctrehdeibility percentage. This component communicates tobodyMap the list of gestures to be displayed according to the filters. It has all the attributes related to the filters (those active, the total list of filters). As far as methods are concerned, these are mainly methods aimed at adapting its display according to the display of the other two components (if these are closed, you can increase the size of DataFilter, which is achieved with returnClass, mainClass). • The last componentIt,emDisplay, takes care of displaying the gesture that has been selected on thbeodyMap or in the list of gestures. It therefore depends on these 2 components in order to obtain the gesture selected by the user. This element is used to display the advanced details of the chosen gesture. The main attribute of this element is the user selected gesture (item). Shows gesture information such as its name, name of origin GES, authors, URL of study, year of publicatciroendi,bility, etc. (see Fig.3).
To provide quantified, peer-reviewed gestures to inform the design of gesture-based user interfaces, it is important that each stored gestures includes relevant information to become efective. In some other references, recommended gestures could be based on the personal opinions of a few experts, do not provide any reference to support them or any empirical evidence to backup their application, do not provide any indication as to whether a particular gesture represents a consensus of researchers or a large agreement among participants, do not give any information about the relative importance of individual GES. To this end, a numerical measure was proposed to quantify tchredibility of consensus gestures ofered by a GES. It has reflected the essential criteria for a GES considering: (a) the length of the study (the number of pages, e.g.,a poster of 4 pages is diferent from a paper of 25 pages), (b) the expertise of the authors in gesture researec.gh.,(, how many papers they published on topics related to gestures,e.g.,7 for an author based on Google Scholar entries, (c) the venue where the GES was published (for which we useScimago’s journal rankinign terms of Q1/Q2/Q3/Q4/none categories anCdORE Rankings Portaflor conferences in terms of A * / A / B / C / D), (d) the number of participants involved in the stue.dgy.,,(a GES with 5 participants is assumed to have a lower validity than a GES with >30 participants), and (e) their diversity in age, we use the standard deviation of participants’ ae.gge.,s, S(D ages = 5 when reported in the GES. It combines (a), (b), (c), (d) and (e) in one singlSetrength of evidence measure, defined as follows:
( )2 + ( )2 + ( )2 + ( )2 + ( )2 (GES) = 5 (1) where: • A = the typical limit of page numbers for a full paper at the major HCI conferee.gn.,c1e0s ( pages), so A=10. If A >= 10, then a/A is bounded to 1, so that each component of the sum above is between 0 and 1. • B = the total number of articles published by all authors of the GES study, and b is the total number of gesture articles published by all authors of the GES study. • C = 5 and we encode Q1 = 5, Q2 = 4, Q3 = 3, Q4 = 2, and other = 1 (low strength of evidence based on the estimated quality of peer review). • D = 20 (the typical number of participants in GES studies; this value should result from the analysis of the appendix where the number of participants is discussed). If d >= 20, then d/D is limited to 1, so that each component of the sum above is between 0 and 1. • E = the standard deviation of participants’ ages extracted or coem.gp.u,,tEe=d4,.15.
For example, Fig.3 displays a list of gestures coming from diferent GES, ”Yes gesture” has a strength of evidence o=f0.54 , and Bend up bed down has a strength of evidence=0o.f6 .
Through my participation in the Doctoral Consortium at EICS22, we were recommended to foster qualitative data, to propose a legacy classification of gesture names that are the same but difer across studies. Inspired by this, we incorporatSeanakey diagramusing the”chartjs-chartsankey”library to show the diferent relationships that exist between gestures and referents (See Fig. 4).
Validation ofGEStory. To validatGeEStory we carry out tests with a group of twelve volunteer participants (7 Males, 4 Females, 1 not specify, aged1f7rtoom79 years). Recordings of these interviews were made to allow calculation of success/failure rates per task and completion of the UEQ+ questionnaire.
In order to perform our tests, I have drawn up a list of 4 actions that the ”testers” are required to perform. I decided on them based on the changes made to the platform. This allows me to see how the user behaves in front of the diferent navigation tools (the selection menu, the navigation bar, the “suggest a GES” button, etc), Below is the list of actions: 1. Look for an iconic dynamic type gesture of 2018 2. Submit new gesture data 3. Look for the ”Move hand UP” from 2015, to be performed with the arm. 4. Find the information containing the name of the tool on which the prototype was built Following the UEQ+ analysis, figur5eshows that the two most important parameters for test participants are dependability (2.42/3) and eficiency (2.33/3). Looking at the “ratings” assigned by the testers to the organization oGf EtShteory platform, we can see in figure6 that these two parameters are among the top five rated (with ours of 0.63 and 0.58 respectively). 3 2 1 0
In Fig. 6, The best ranked parameter of the participants is that relating to the intuitive use of theGEStory platform (1.04/3). This allows us to link this last criterion with the hypotheses we have developed above. Indeed, we can quite easily say that we have succeeded in suficiently reducing the workload of the platform (which seemed very high to us on the initial version of the project) so that the user can handle this tool without needing a very extensive training. We can also highlight the fact that the accounting of the interface is quite suitable for it to allow users to carry out their various tasks. This allows us to draw a parallel with the success rate of the tasks we asked them to perform which is, for each task, at least, grea7t5e%r than success.The utility is a parameter of the UEQ+ analysis that had the most negative result (-0.98). If we rely on the given definition, this parameter represents the fact that the use of the product brings benefits to the user.
Fig. 7 shows the result of task success/failure rate. In general, the participants were able to perform the various tasks that were asked of them. The least successful task being number 1 (9 successes, 2 successes with our help, and 1 failure, see F7iag). During our tests, we were able to observe that the failure of the task was often explained by a phase, on the part of the user, of ”taking control of the platform”. Most of the participants took the time to discover the diferent search tools (the search engine and the search criteria system) and sometimes did not understand how they work directly (despite a platform presentation phase). performed before the various tests).
By looking at the average resolution time of each task, we can also notice that task No. 1 is the one that took the longest to complete (which could also be justified by this phase of ”getting started with the tool”). The average resolution times being for task No. 1: 73 seconds, task No. 2: 27 seconds, task No. 3: 61 seconds, task No. 4: 24 seconds.
These task resolution times also allow us to observe the fact that the tasks requiring the use of the search tools present on the site (task No. 1 and No. 3) required longer resolution times than the two others. As a reminder, the first task being: Look for an iconic dynamic type gesture from 2018, and the third: Look for the ‘Move hand UP’ from 2015, to be performed with the arm.
Transition toGESistant. A transition should be ensured betwGeEenStory andGESistant, a software aimed at assisting the experimenter to conduct a GES in a way that is distributed in time (stages are asynchronous) and space (participants are contributing remotely, self-assisted and without any constraint) structured into 6 stages: define a study, conduct a study, classify gestures, measure gestures, discuss gestures, and export gestures. When a quGeErSytionry does not lead to a compelling set of appropriate gestures, the parameters of the query should be transferred to the “define a study” stageGEinSistant in order to match the suggested GES criteria.
Based on the research methodology (F1i)g, .the first SLR has been completed and its results are published2[]. The second SLR of gesture characteristics has been completed in terms of research and are momentarily stored as on-line spreadsheets. Their results are under analysis. Other GES have been conducted and published. Several others have been conducted, but not yet analyzed, such as a GES for a haptic devvisc.ewithout it.
The GEStory prototype was created based on its domain model 2(F)iwg.hose classes include Body part, Device, Gesture, Environment, Participant, Study, Referent Classification (FunctionSubType and FunctionType). The relationships between device and gesture, body part and gesture, gesture and referent, participant and study, study and gesture, gesture and environment are also considered. CurrenGtElSyt,ory [27] is considered an interactive design space [28], whose classes are dimensions of investigation. Checking or unchecking the values of each dimension will automatically result in the display of GES satisfying these criteria and their consensus gestures (Fig3.). Each gesture is displayed according to a textual representation, a textual description, a picture, JSON-based formal definition based on an Extended Backus-Naur Form (EBNF) grammar with transformations betwe2e6n]. [By means of a Sankey diagram (Fig. 4) the relationship between the classified gestures and the classified referents is shown; this diagram shows the distribution of the preference of the users of a gesture to perform a referent.
A validation of the current versioGnEoSftory was carried out in which 12 participants evaluated ”Attractiveness”, ”Eficiency”, ”Trust”, ”Dependability”, ”Adaptability”, ”Usefulness”, ”Value”, ”Intuitive use” and ”Quality of content” with the UEQ+ questionna5iraen(dFiFgi.g.6) and performed 4 tasks for which the time to perform and the success/failure rate were calculated (Fig. 7).
It is developing the transition frGoEmStory toGESistant, this will allow the user to export theGEStory informatione(.g.,GES result for replication, configuration parameters for a new GES, etc.), This will allow experimenters to have a preload for a newGEGSEisSt.ant will allow the study to be carried out remotely and asynchronously.
The author of this paper acknowledges the support of SIGCHI Gary Marsden Travel Awards 2022 for the support given for the trip aAnFdIHM for the invitation of Doctoral Consortium EICS2022. Review, Multimodal Technologies and Interaction 2 (2018) 65–. hUtRtLp:s://www.mdpi. com/2414-4088/2/4/65. doi:https://doi.org/10.3390/mti2040065. [12] P. Koutsabasis, P. Vogiatzidakis, Empirical research in mid-air interaction: A systematic review, International Journal of Human-Computer Interaction (2019) 1–22. URL: https://doi.org/10.1080/10447318.2019.157235.2doi:10.1080/10447318.2019.1572352. arXiv:https://doi.org/10.1080/10447318.2019.1572352. [13] N. Magrofuoco, P. Roselli, J. Vanderdonckt, J. L. Pérez-Medina, R.-D. Vatavu, Gestman: A cloud-based tool for stroke-gesture datasets, in: Proceedings of the ACM SIGCHI Symposium on Engineering Interactive Computing Systems, EICS ’19, Association for Computing Machinery, New York, NY, USA, 2019, pp. 1–6. URLh: ttps://doi.org/10.1145/ 3319499.3328227. doi:10.1145/3319499.3328227. [14] L. Anthony, R.-D. Vatavu, J. O. Wobbrock, Understanding the consistency of users’ pen and finger stroke gesture articulation, in: Proceedings of Graphics Interface 2013, GI ’13, Canadian Information Processing Society, CAN, 2013, p. 87–94. [15] L. A. Leiva, R.-D. Vatavu, D. Martín-Albo, R. Plamondon, Omnis praedictio: Estimating the full spectrum of human performance with stroke gestures, International Journal of Human-Computer Studies 142 (2020) 102466. URLh:ttps://www.sciencedirect.com/science/ article/pii/S107158192030068.9doi:https://doi.org/10.1016/j.ijhcs.2020.102466. [16] H. Dang, D. Buschek, Gesturemap: Supporting visual analytics and quantitative analysis of motion elicitation data by learning 2d embeddings, in: Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems, CHI ’21, Association for Computing Machinery, New York, NY, USA, 2021, pp. 1–12. URL:https://doi.org/10.1145/3411764. 3445765. doi:10.1145/3411764.3445765. [17] R.-D. Vatavu, L.-B. Bilius, GestuRING: A Web-Based Tool for Designing Gesture Input with Rings, Ring-Like, and Ring-Ready Devices, Association for Computing Machinery, New York, NY, USA, 2021, p. 710–723. URL:https://doi.org/10.1145/3472749.347478.0 [18] B.-F. Gheran, J. Vanderdonckt, R.-D. Vatavu, Gestures for smart rings: Empirical results, insights, and design implications, in: Proceedings of the 2018 Designing Interactive Systems Conference, DIS ’18, Association for Computing Machinery, New York, NY, USA, 2018, p. 623–635. URL: https://doi.org/10.1145/3196709.319674.1doi:10.1145/3196709.3196741. [19] A. Liberati, D. G. Altman, J. Tetzlaf, C. Mulrow, P. C. Gøtzsche, J. P. A. Ioannidis, M. Clarke, P. J. Devereaux, J. Kleijnen, D. Moher, The prisma statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration, PLoS Medicine 6 (2009) 1–22. UhRtLt:ps://www.ncbi.nlm.nih.gov/ pubmed/19621070. doi:10.1371/journal.pmed.1000100. [20] D. R. Lenorovitz, M. D. Phillips, R. Ardrey, G. V. Kloster, A taxonomic approach to characterizing human-computer interaction., in: G. Salvendy (Ed.), Human-Computer Interaction., Elsevier Science Publishers, Amsterdam, 1984, pp. 111–116. [21] R. Aigner, D. Wigdor, H. Benko, M. Haller, D. Lindbauer, A. Ion, S. Zhao, J. T.
K. V. Koh, Understanding Mid-Air Hand Gestures: A Study of Human Preferences in Usage of Gesture Types for HCI, Technical Report MSR-TR-2012-111, Microsoft Research, 2012. URL:https://www.microsoft.com/en-us/research/wp-content/uploads/2016/ 02/GesturesTR-20121107-RoA.pd.f [22] T. Piumsomboon, A. Clark, M. Billinghurst, A. Cockburn, User-defined gestures for augmented reality, in: CHI ’13 Extended Abstracts on Human Factors in Computing Systems, CHI EA ’13, ACM, New York, NY, USA, 2013, pp. 955–960. URL:http://doi.acm. org/10.1145/2468356.2468527. doi:10.1145/2468356.2468527. [23] M. Obaid, M. Häring, F. Kistler, R. Bühling, E. André, User-defined body gestures for navigational control of a humanoid robot, in: S. S. Ge, O. Khatib, J.-J. Cabibihan, R. Simmons, M.-A. Williams (Eds.), Social Robotics, Lecture Notes in Computer Science, Springer, Berlin, Heidelberg, 2012, pp. 367–377. URLh:ttps://doi.org/10.1007/978-3-642-34103-8_3.7 doi:10.1007/978-3-642-34103-8_37. [24] E. McAweeney, H. Zhang, M. Nebeling, User-driven design principles for gesture representations, in: Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems, CHI ’18, Association for Computing Machinery, New York, NY, USA, 2018, pp. 1–13. URL: https://doi.org/10.1145/3173574.317412.1doi:10.1145/3173574.3174121. [25] M. R. Morris, A. Danielescu, S. Drucker, D. Fisher, B. Lee, m. c. schraefel, J. O. Wobbrock, Reducing legacy bias in gesture elicitation studies, Interactions 21 (2014) 40–45. URL: https://doi.org/10.1145/259168.9doi:10.1145/2591689. [26] N. Aquino, J. Vanderdonckt, O. Pastor, Transformation templates: Adding flexibility to model-driven engineering of user interfaces, in: Proceedings of the 2010 ACM Symposium on Applied Computing, SAC ’10, Association for Computing Machinery, New York, NY, USA, 2010, p. 1195–1202. URL: https://doi.org/10.1145/1774088.177434.0doi:10.1145/ 1774088.1774340. [27] B.-F. Gheran, S. Villarreal-Narvaez, R.-D. Vatavu, J. Vanderdonckt, Repliges and gestory: Visual tools for systematizing and consolidating knowledge on user-defined gestures, in: Proceedings of the 2022 International Conference on Advanced Visual Interfaces, AVI 2022, Association for Computing Machinery, New York, NY, USA, 2022, pp. 1–9. URL: https://doi.org/10.1145/3531073.353111.2doi:10.1145/3531073.3531112. [28] F. Heller, K. Todi, K. Luyten, An Interactive Design Space for Wearable Displays, Association for Computing Machinery, New York, NY, USA, 2021, p. 14. URLh:ttps://doi.org/10.1145/ 3447526.3472034. [29] J. Vanderdonckt, Accessing guidelines information with sierra, in: Proc. of IFIP TC13 Int.
Conf. on Human-Computer Interaction, INTERACT ’95, Chapman & Hall, London, 1995, pp. 311–316. [30] A. Beirekdar, J. Vanderdonckt, M. Noirhomme-Fraiture, A Framework and a Language for Usability Automatic Evaluation of Web Sites by Static Analysis of HTML Source Code, Springer Netherlands, Dordrecht, 2002, pp. 337–348. UhRtLt:ps://doi.org/10.1007/ 978-94-010-0421-3_29. doi:10.1007/978-94-010-0421-3_29.