BieColl - Bielefeld Electronic Collections

Abbott, William Welby

Faisal, Aldo Ahmed

Ultra low-cost 3D gaze estimation: an intuitive 
high information throughput complement to 
direct Brain-Machine-Interfaces  
William Welby Abbott and Aldo Ahmed Faisal
The advancement of Brain Machine Interface (BMI) technology for controlling neu-
romotor prosthetic devices holds the hope to restore vital degrees of independence 
to patients with motor disorders, improving quality of life. Unfortunately, emerging 
rehabilitative methods come at considerable clinical and post-clinical operational 
cost, beyond the means of the majority of patients (Shih et al. 2012). We present an 
ultra-low cost alternative: using eye-tracking. Monitoring eye movement provides 
a feasible alternative to traditional BMIs because the ability to control eye-move-
ments can be retained in cases of severe traumas or pathologies in which all other 
motor functions are lost (Kaminski et al. 2002; Kaminski et al. 1992). Based on 
disease statistics, we find that within the EU alone, there were over 16 million peo-
ple in 2005 (3.2% of the population) with disabilities that would benefit from such 
gaze based communication and control systems (Jordansen et al. 2005). Despite 
this, eye tracking is not widely used as control interface for movement impaired 
patients due to high cost, poor signal interpretation and lack of control flexibility.  
We propose that tracking the gaze position in 3D rather than 2D (as is used in 
most gaze interaction methods (Majaranta et al., 2011)) provides a considerably 
richer signal for neuroprosthetic control by allowing direct interaction with the en-
vironment rather than via computer displays. We demonstrate that by using mass-
produced video-game hardware that an ultra-low cost binocular eye-tracker with 
comparable performance to commercial systems more than 800 times as expensive 
is possible (Abbott & Faisal 2012). Our head-mounted system has 30 USD material 
costs and operates at over 120 Hz sampling rate with a 0.5-1 degree of visual angle 
resolution. We perform 2D and 3D gaze estimation, controlling a real-time volu-
metric cursor, essential for driving complex user interfaces. Our approach yields an 
information throughput of 43 bits/s, more than ten times that of invasive and semi-
invasive BMIs that are vastly more expensive.  
William Welby Abbott  
    Imperial College London, Dept. Bioengineering, UK 
Aldo Ahmed Faisal 
  Imperial College London, Dept. of Computing and Dept. Bioengineering, UK 
DOI: 10.2390/biecoll-saga2013_14
In Pfeiffer and Essig (2013): Proceedings of the First International Workshop on Solutions for Automatic Gaze-Data Analysis 2013 
(SAGA 2013). DOI: 10.2390/biecoll-saga2013_0
Ultra low-cost 3D gaze estimation  
In contrast, BMI information rates from direct recording of neuronal activity are 
ultimately constrained by noise in the recording systems and the nervous system 
itself (Faisal et al. 2008). In particular physical noise sources inside central neurons 
(Faisal et al. 2002; Faisal 2009) and peripheral axons (Faisal et al. 2005) will limit 
decoding performance from limited numbers of independent neuronal sources. 
Thus, to compensate for noise, BMI signal decoders have to observe signals for 
longer periods of time, thereby increasing response latencies for direct BMIs.  While 
these issues will be ameliorated by the steady progress of sensor quality and density, 
eye movements already offer a highly accurate, low-latency (and low cost) read out. 
This is because the brain has already evolved to minimise the role of noise and 
delays in eye movements, which form an aggregated output of the nervous system. 
The leap in readout performance (in terms of readout performance and latency) en-
ables closed-loop real-time control of rehabilitative and domotic devices beyond 
what is achievable by current BMIs: e.g. it was estimated that powered wheelchair 
control requires, on average, 15.3 bits/second and full-finger hand prosthetics re-
quire 54.2 bits/second (Tonet et al. 2008). We have demonstrated how an ultra-low 
cost, non-invasive eye-tracking approach can form the basis of a real-time control 
interface for rehabilitative devices – making it a low-cost complement or alternative 
to existing BMI technologies.  
References 
1. Abbott, W.W. & Faisal, A.A., 2012. Ultra-low-cost 3D gaze estimation: an intuitive high in-
formation throughput compliment to direct brain–machine interfaces. Journal of Neural Engi-
neering, 9(4), p.046016.
2. Faisal, A.A., 2009. Stochastic Simulation of Neurons, Axons, and Action Potentials. In C.
Laing & G. J. Lord, eds. Stochastic Methods in Neuroscience. Oxford University Press, pp.
297–343.
3. Faisal, A.A., Laughlin, S.B. & White, J.A., 2002. How reliable is the connectivity in cortical
neural networks? In Neural Networks, 2002. IJCNN’02. Proceedings of the 2002 International
Joint Conference on. pp. 1661–1666.
4. Faisal, A.A., Selen, L.P.J. & Wolpert, D.M., 2008. Noise in the nervous system. Nature Re-
views Neuroscience, 9(4), pp.292–303.
5. Faisal, A.A., White, J.A. & Laughlin, S.B., 2005. Ion-Channel Noise Places Limits on the Min-
iaturization of the Brain’s Wiring. Current Biology, 15(12), pp.1143–1149.
6. Jordansen, I. et al., 2005. Report on a market study and demographics of user population
(cogain) ist-2003-511598: Deliverable 7.2),
7. Kaminski, H.J. et al., 2002. Differential Susceptibility of the Ocular Motor System to Disease.
Annals of the New York Academy of Sciences, 956(1), pp.42–54.
8. Kaminski, H.J. et al., 1992. Extraocular muscles are spared in advanced duchenne dystrophy.
Annals of Neurology, 32(4), pp.586–588.
9. Majaranta, P., Aoki, H., Donegan, M., Hansen, D. W., Hansen, J. P., Hyrskykari, A., & Räihä,
K.-J. (2011). Gaze Interaction and Applications of Eye Tracking. IGI Global.
10. Shih, J.J., Krusienski, D.J. & Wolpaw, J.R., 2012. Brain-Computer Interfaces in Medicine.
Mayo Clinic Proceedings, 87(3), pp.268–279.
11. Tonet, O. et al., 2008. Defining brain-machine interface applications by matching interface
performance with device requirements. Journal of Neuroscience Methods, 167(1), pp.91–104.


English

Ultra low-cost 3D gaze estimation: an intuitive high information throughput complement to direct Brain-Machine-Interfaces

Uslucan, Haci-Halil

Psychologische Bedingungen des Bildungserfolges von Migranten

Schindler, Kirsten

Texte beurteilen - Feedback geben. Kompetenzen für Lehramtsstudierende

Broschinski, Christoph

Vompras, Johanna

Kollaborative Arbeitsplattform (VFU) und Tools für SFB882

Paletta, Lucas

Santner, Katrin

Hofmann, Albert

Thallinger, Georg

3D Gaze Recovery in Large Environments 
Using Visual SLAM  
Lucas Paletta, Katrin Santner, Albert Hofmann, Georg Thallinger 
DIGITAL – Institute for Information and Communication Technology 
JOANNEUM RESEARCH Forschungsgesellschaft mbH, 8010 Graz, Austria 
1 Introduction 
This work describes a multi-component vision system that enables pervasive map-
ping of human attention. The key contribution is that our methodology enables full 
3D recovery of the gaze pointer, human view frustum and associated human cen-
tered measurements directly into an automatically computed 3D model. We apply 
RGB-D SLAM and descriptor matching methodologies for the 3D modeling, lo-
calization and fully automated annotation of ROIs (regions of interest) within the 
acquired 3D model. This methodology enables fully automated processing of hu-
man attention, without artificial landmarks, in indoor natural environments. 
2. Implementation
This work presents a computer vision system methodology that, firstly, enables 
to precisely estimate the 3D position and orientation of human view frustum and 
gaze [1] and from this enables to precisely analyze human attention in the context 
of the semantics of the local environment (objects [10], signs, scenes, etc.). Second-
ly, the work describes how ROIs (regions of interest) are automatically mapped 
from a reference video into the model and from this prevents from state-of-the-art 
laborious manual labeling of tens / hundreds of hours of eye tracking video data. 
This provides a scaling up of nowadays still small sketched attention studies. With 
the presented methodology, extended natural environments, such as shop floor de-
partments, analysis of navigation guidance, and human-robot interaction, can be 
studied first time in large scale, statistically significant usability studies.  
For a spatio-temporal analysis of human attention in the 3D environment, we first-
ly build a spatial reference in terms of a three-dimensional model of the environ-
ment using RGB-D SLAM methodology [2]. Secondly, the user’s view is gathered 
with eye tracking glasses (ETG) within the environment and localized from ex-
tracted local descriptors [3]. Then ROIs are marked on imagery and automatically 
detected in video and further mapped into the 3D model. Finally, the distribution 
of saliency onto the 3D environment is computed for further human attention 
analysis, such as, evaluation of the attention mapping with respect to object and 
scene awareness. Saliency information can be aggregated and further evaluated in 
the frame of user behaviors of interest. The performance evaluation of the present-
DOI: 10.2390/biecoll-saga2013_8
In Pfeiffer and Essig (2013): Proceedings of the First International Workshop on Solutions for Automatic Gaze-Data Analysis 2013 
(SAGA 2013). DOI: 10.2390/biecoll-saga2013_0
3D Gaze Recovery in Large Environments Using Visual SLAM 
ed methodology firstly refers to results from a dedicated test environment [4] 
demonstrating very low projection errors, enabling to capture attention on daily 
objects and activities (package logos, cups, books, pencils).  
3 Gaze recovery in 3D without artificial markers 
Human Attention Analysis in 3D. 3D information recovery of human gaze has in 
principle been targeted by Munn et al. [2] who introduced monocular eye-tracking 
and triangulation of 2D gaze positions of subsequent key video frames, obtaining 
observer position and gaze pointer in 3D with angular errors of ≈3.8°. Pirri et al. 
[3] achieved accuracy indoors about ≈3.6 cm at 2 m distance to the target com-
pared to our ≈0.9 cm [1]. Previous attempts focused on single 3D point recovery. 
Our approach maps fixation within a 3D environment model with the possibility 
of real-time tracking of attention with mass marketed eye-tracking hardware. 
Visual Map Building and Camera Pose Estimation. For realistic environ-
ment modeling we make use of an RGB-D sensor providing per pixel color and 
depth information at high frame rates. Our environment consists of a sparse point-
cloud, where each landmark [4] is attached for data association during pose track-
ing. Estimated camera poses are stored in a 6DOF manner. Incremental camera 
pose tracking assuming an already existing map is done by keypoint matching fol-
lowed by a least-square optimization routine minimizing the reprojection.  
Densely Textured Surface Generation. For realistic environment visualiza-
tion, user interaction and subsequent human attention analysis, a dense, textured 
model of the environment is constructed. Depth images are integrated into a 3D 
occupancy grid [5] using the previously corrected camera pose estimates.  
3D Gaze Recovery from Monocular Localization. To estimate the proband’s 
pose, SIFT keypoints are extracted from ETG video frames and a full 6DOF pose 
is estimated using the perspective n-Point algorithm [6].  
Automated 3D Annotation of Regions of Interest. Annotation of ROIs in 2D 
or even 3D information usually causes a process of massive manual interaction. In 
order to map objects of interests, such as, logos, package covers, etc. into the 3D 
model, we first use logo detection in the high resolution scanning video to search 
for occurrences of predefined reference appearances, using vocabulary trees [4].  
Semantic Mapping of Attention. The automatic detection of ROIs in 3D ena-
bles statistical evaluations, such as on ROIs called AOI hit, which states for a raw 
sample or a fixation that its coordinate value is inside the ROI [7]. From this, the 
dwell time distribution for ROIs can be plotted over all participants, and some of 
the captured fixations are related to human object recognition which is known to 
trigger from 100 ms of observation / fixation [8].  
Lucas Paletta, Katrin Santner, Albert Hofmann, Georg Thallinger 
4 Experimental results 
Eye Tracking Device. The mass marketed SMI™ eye-tracking measure the 
gaze pointer for both eyes with 30 Hz. The gaze pointer accuracy of 0.5°–1.0° and 
a tracking range of 80°/60° horizontal/vertical assure a precise localization of the 
human’s gaze in the HD 1280x960 scene video with 24fps. We recorded data on a 
shop floor covering an area of about 8x20m² (Figure 1). We captured 2366 RGB-D 
images, reconstructed the model from 41700 natural visual landmarks.   
5 Conclusions and future work 
We present a complete system for (i) wearable data capturing, (ii) automated 
3D modeling, (iii) automated recovery of human pose and gaze, and (iv) automat-
ed ROI based semantic interpretation of human attention. The presented system is 
a significant step towards a mobile mapping framework [9] for quantitative analy-
sis of human attention measures [7,10] in natural environments (Figure 2). Future 
work will focus on improved tracking of the human pose across image blur and 
uncharted areas as well as study human factors in the frame of stress and emotion 
in the context of the 3D space. 
(a) (b)
(c) 
Figure 1. Hardware (a) for the 3D model building process (Kinect and HD camera), (b) 
study with packages, (c) 3D environment model, a large shop floor. 
3D Gaze Recovery in Large Environments Using Visual SLAM 
ACKNOWLEDGMENTS: The research leading to these results has received funding from the 
European Community's Seventh Framework Programme (FP7/2007-2013) under grant agree-
ment No. 288587 (MASELTOV) as well as from the Austrian FFG via contracts No. 832045 
(FACTS) and No. 836270 (EVES), and by the Provinc. Gov. of Styria (NeoAttrakt). We kindly 
thank INTERSPAR Graz and CITYPARK GmbH for the permission to capture the data.  
References 
1. Paletta, L., Santner, K., Fritz, G., Mayer, H., & Schrammel, J.: 3D Attention: Measurement of
Visual Saliency Using Eye Tracking Glasses, Proc. CHI 2013.
2. Munn, S. M., & Pelz J. B.: 3D point-of-regard, position and head orientation from a portable 
monocular video-based eye tracker. Proc. ETRA 2008, pp. 181-188.
3. Pirri, F., Pizzoli, M., & Rudi, A.: A general method for the point of regard estimation in 3D
space. Proc. CVPR 2011, pp. 921-928.
4. Nistér, D. & Stewénius, H.: Scalable Recogn. with a Vocabulary Tree, Proc. CVPR 2006.
5. Marks, T. K., Howard, A., Bajracharya, M., Cottrell, G. W. & Matthies, L.: Gamma-SLAM:
Using stereo vision and variance grid maps for SLAM, Proc. ICRA 2008.
6. Lepetit V., Moreno-Noguer F. and Fua P.: EPnP: An Accurate O(n) Solution to the PnP Prob-
lem, International Journal of Computer Vision, pp. 155-166, 2009.
7. Holmqvist, K., Nyström, M., Andersson, R., Dewhurst, R., Jarodzka, H., and van de Weijler,
J.: Eye Tracking, Oxford University Press, 2011, pp. 187.
8. Grill-Spector, K. and Sayres, R. Object Recognition: Insights From Advances in fMRI Meth-
ods, Current Direct. Psychol. Science, Vol. 17, No. 2. 2008, pp. 73-79.
9. Paletta, L., Santner, K., Fritz, G., Hofmann, A., Lodron, G., Thallinger, G., and Mayer, H.
FACTS - A Computer Vision System for 3D Recovery and Semantic Mapping of Human
Factors, Proc. ICVS 2013, Springer-Verlag, LNCS 7963, pp. 62-72.
10. Fritz, G., Seifert, C., Paletta, L., and Bischof, H. (2005). Attentive object detection using an 
inform. theoret. saliency measure, Proc. WAPCV 2004, Springer-Verlag, LNCS 3368.
Figure 2. Mapping of user saliency onto the acquired 3D model and automated recovery of 
the trajectory of ETG camera positions (spheres), as well as recovery of frustum (green 
lines) and gaze pointer (blue). 


3D Gaze Recovery in Large Environments Using Visual SLAM

Petersen, Inger

German

Entwicklung schriftlicher Argumentationskompetenz bei ein- und mehrsprachigen Oberstufenschüler/-innen und Studierenden

Lösch, Mathias

.
Universität Bielefeld
Das BASE-Netzwerk
Services und Applikationen im Umfeld der Bielefeld Academic Search
Engine
Mathias Lösch
Universitätsbibliothek Bielefeld – LibTec
Kolloquium Wissensinfrastruktur 17.5.2013
Das BASE-Netzwerk M. Lösch 1 / 35
.
Universität Bielefeld
.
BASE
Voraussetzungen Services Ausblick Zusammenfassung Literatur
∼45M Dokumente
> 2.500 Quellen
Das BASE-Netzwerk M. Lösch 2 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 3 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 4 / 35
.
Universität Bielefeld
.
Vollständigkeit
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Quellenakquise
Repository-Registries
Open Archives Registry
OpenDOAR
ROAR
University of Illinois Registry
OJS Registry
Bekanntheit in der Community
Vorschlagsformular
Das BASE-Netzwerk M. Lösch 5 / 35
.
Universität Bielefeld
.
Datenqualität // Digital Curation
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Kuratierung der Datenquellen
Herkunftsland
Kontinent
Typisierung
Repository-Software
Open-Access-Anteil
Das BASE-Netzwerk M. Lösch 6 / 35
.
Universität Bielefeld
.
Datenqualität // Metadatenstandards
Voraussetzungen Services Ausblick Zusammenfassung Literatur
OAI-PMH-Expertise
Feedback, Beratung
Validator
http://oval.base-search.net/
Guidelines
DINI
DRIVER
OpenAIRE
Das BASE-Netzwerk M. Lösch 7 / 35
.
Universität Bielefeld
.
Datenqualität // Normalisierungen
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Normalisierungen der Metadatenfelder
Sprache “eng”,“en”, “en_US”, “English” → “eng”
Dokumenttyp “article”, “doc-type:article” → “0001”
Publikationsjahr “2004”, “2004-10-20”, “April 2004” → “2004”
Automatische Klassifikation
“The notion of a Hilbert space…” → DDC 510 Mathematics
Das BASE-Netzwerk M. Lösch 8 / 35
.
Universität Bielefeld
.
Datenqualität // Normalisierungen
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Normalisierungen der Metadatenfelder
Sprache “eng”,“en”, “en_US”, “English” → “eng”
Dokumenttyp “article”, “doc-type:article” → “0001”
Publikationsjahr “2004”, “2004-10-20”, “April 2004” → “2004”
Automatische Klassifikation
“The notion of a Hilbert space…” → DDC 510 Mathematics
Das BASE-Netzwerk M. Lösch 8 / 35
.
Universität Bielefeld
.
Schnittstellen // BASE-Architektur
Voraussetzungen Services Ausblick Zusammenfassung Literatur
.
.Applikationen .
Frontend
.
Service1
.
Service2
.
Service3
.Index ..Index1 ..Index2 ..Index3 ..Index4
.Indexierung . .Indexierung . .
.Preprocessing .Klassifikation .Normalisierung . .
.Aggregation .Harvesting . . .
Das BASE-Netzwerk M. Lösch 9 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 10 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 11 / 35
.
Universität Bielefeld
.
Suchschnittstelle
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Einbindung der BASE-Suche in
externe Anwendungen
z.B. Bibliothekskataloge:
Katalog.plus UB Bielefeld
Katalog d. UB Heidelberg
Metasuchmaschinen
Metager
Karlsruher Virtueller Katalog
(KVK)
Das BASE-Netzwerk M. Lösch 12 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 13 / 35
.
Universität Bielefeld
.
Rsync-Schnittstelle
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Gesamte BASE-Datenbasis als Download
Anwendungen
AuthorClaim
Autoren „claimen“ ihre
Publikationen (u.a.) aus dem
BASE Pool
http://authorclaim.org/
High North Research Documents
BASE-Dokumente aus dem
Gebiet der Polarforschung
http://highnorth.uit.no/
Das BASE-Netzwerk M. Lösch 14 / 35
.
Universität Bielefeld
.
Rsync-Schnittstelle
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Gesamte BASE-Datenbasis als Download
Anwendungen
AuthorClaim
Autoren „claimen“ ihre
Publikationen (u.a.) aus dem
BASE Pool
http://authorclaim.org/
High North Research Documents
BASE-Dokumente aus dem
Gebiet der Polarforschung
http://highnorth.uit.no/
Das BASE-Netzwerk M. Lösch 14 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 15 / 35
.
Universität Bielefeld
.
OAI-Schnittstelle // http://oai.base-search.net/
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Ziele
Zugriff auf Daten-Ausschnitte aus BASE
Fachlich (DDC-basiert)
geographisch
nach Dokumentsprache
nach Quelle
Kombinationen
Protokoll: OAI-PMH
Standard-Client-Software vorhanden
Breite Akzeptanz bei der Zielgruppe
Das BASE-Netzwerk M. Lösch 16 / 35
.
Universität Bielefeld
.
OAI-Schnittstelle // Implementierung
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Implementierung
Erweiterung des OAI-PMH-Protokolls um „Dynamische OAI-Sets“
(inspiriert durch DataCite)
Set-Spezifikation mit der Solr-Query-Syntax:
set=ddc:330 vs.
set=deweyfull:33* AND language:(eng OR ger)
Das BASE-Netzwerk M. Lösch 17 / 35
.
Universität Bielefeld
.
OAI-Schnittstelle // Spezifizierung von Ausschnitten
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Beispiel: Spezifizierung fachlicher Ausschnitte
Wirtschaftswissenschaften: deweyfull:33*
Wirtschaftswissenschaften und BWL: deweyfull:33* AND
deweyfull:65*
Physik und Astronomie: deweyfull:53* AND deweyfull:52*
Das BASE-Netzwerk M. Lösch 18 / 35
.
Universität Bielefeld
.
OAI-Schnittstelle // Anwender
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Anwender
Virtuelle Fachbibliotheken
EconBiz: Virtuelle Fachbibliothek Wirtschaft (ZBW)
Fachportal Pädagogik: Erziehungswissenschaft (DIPF)
OpenAIREplus
Trainingsdaten für Dokumentklassifikation und -clustering
Europeana Cloud
Fachbereich „Social Sciences & Humanities“
deweyfull:* AND -deweyfull:0* AND -deweyfull:5* AND
-deweyfull:6* AND continent:ceu
Das BASE-Netzwerk M. Lösch 19 / 35
.
Universität Bielefeld
.
OAI-Schnittstelle // Anwender
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Anwender
Virtuelle Fachbibliotheken
EconBiz: Virtuelle Fachbibliothek Wirtschaft (ZBW)
Fachportal Pädagogik: Erziehungswissenschaft (DIPF)
OpenAIREplus
Trainingsdaten für Dokumentklassifikation und -clustering
Europeana Cloud
Fachbereich „Social Sciences & Humanities“
deweyfull:* AND -deweyfull:0* AND -deweyfull:5* AND
-deweyfull:6* AND continent:ceu
Das BASE-Netzwerk M. Lösch 19 / 35
.
Universität Bielefeld
.
OAI-Schnittstelle // Anwender
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Anwender
Virtuelle Fachbibliotheken
EconBiz: Virtuelle Fachbibliothek Wirtschaft (ZBW)
Fachportal Pädagogik: Erziehungswissenschaft (DIPF)
OpenAIREplus
Trainingsdaten für Dokumentklassifikation und -clustering
Europeana Cloud
Fachbereich „Social Sciences & Humanities“
deweyfull:* AND -deweyfull:0* AND -deweyfull:5* AND
-deweyfull:6* AND continent:ceu
Das BASE-Netzwerk M. Lösch 19 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 20 / 35
.
Universität Bielefeld
.
Statistik-Schnittstelle
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Motivation
BASE ist z.Zt. vermutlich das vollständigste Abbild der
internationalen Repositorien-/Open Access-Landschaft
Hohe Datenqualität durch Normalisierungen
Forschungsinteresse:
„Dramatic Growth of Open Access“ (H. Morrison)
Census of Open Access Repositories (Vierkant et al., 2012)
Das BASE-Netzwerk M. Lösch 21 / 35
.
Universität Bielefeld
.
Statistik-Schnittstelle // Funktionen
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Programmatische Abfrage von statistischen Daten aus dem aktuellen
BASE-Index
Tabellarische & graphische Ausgabeformate
Verschiedene Dateiformate
Das BASE-Netzwerk M. Lösch 22 / 35
.
Universität Bielefeld
.
Statistik-Schnittstelle // Plots
Voraussetzungen Services Ausblick Zusammenfassung Literatur
10 größte Quellen
5.52%
4.74%
4.02%
3.73%
2.80%
2.52% 2.49%
1.87% 1.81%
1.30%
Source
Documents
P
u
b
M
e
d
 C
e
n
t
r
al
M
ic
h
ig
an
 U
n
iv
.: 
D
ig
it
al
 L
ib
r
ar
y
Hi
g
h
W
ir
e
 P
r
e
ss
G
al
li
ca
 (
Bi
b
. 
N
at
io
n
al
e
 d
e
 F
r
an
ce
)
D
at
aC
it
e
 M
e
t
ad
at
a 
St
o
r
e
D
O
AJ
 A
r
t
ic
le
s
R
e
P
E
c.
o
r
g
Ar
X
iv
.o
r
g
Ha
t
h
i 
T
r
u
st
 D
ig
it
al
 L
ib
r
ar
y
D
ig
iZ
e
it
sc
h
r
if
t
e
n
0
500,000
1,000,000
1,500,000
2,000,000
2,500,000
3,000,000
Das BASE-Netzwerk M. Lösch 23 / 35
.
Universität Bielefeld
.
Statistik-Schnittstelle // Plots
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Top 10 Sprachen
(normalisiert)
41.46%
33.88%
6.33% 5.29%
2.57% 2.01% 1.70% 1.62% 1.47% 0.55%
Normalized Language
Documents
e
n
g
u
n
kn
o
w
n
fr
e
sp
a
g
e
r
j
pn
po
r
po
l
ch
i
d
u
t
5,000,000
10,000,000
15,000,000
20,000,000
Top 10
Publikationsjahre
5.43%
4.59%
4.26% 4.09% 4.02%
3.76%
3.55% 3.52%
2.55%
1.95%
Year
Documents
20
10
20
11
20
12
20
09
20
08
20
07
20
05
20
06
20
04
20
03
0
500,000
1,000,000
1,500,000
2,000,000
2,500,000
3,000,000
Das BASE-Netzwerk M. Lösch 24 / 35
.
Universität Bielefeld
.
Statistik-Schnittstelle // Plots
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Open Access-Anteil
73.84%
26.16%
Open Access Status
Documents
u
n
kn
o
w
n
O
pe
n
 A
cc
e
ss
10,000,000
20,000,000
30,000,000
40,000,000
Das BASE-Netzwerk M. Lösch 25 / 35
.
Universität Bielefeld
.
Statistik-Schnittstelle // Tabellen
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Top Herkunfts-
länder
(OpenOffice
Spreadsheet)
Das BASE-Netzwerk M. Lösch 26 / 35
.
Universität Bielefeld
.
Statistik-Schnittstelle // Datenformate
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Grafikformate
SVG
PNG
JPG
TIF
PostScript/EPS
PDF
Tabellarische Formate
CSV
ODS
Excel
JSON
YAML
HTML
Das BASE-Netzwerk M. Lösch 27 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 28 / 35
.
Universität Bielefeld
.
Klassifikationsschnittstelle // http://clfapi.base-search.net/doc/
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Bereitstellung der Automatischen DDC-Klassifikation für externe
Applikationen
Nutzer:
OA-Netzwerk (Kindling und Schirmbacher, 2013)
PUB (DDC-Vorschlagsfunktion)
Das BASE-Netzwerk M. Lösch 29 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 30 / 35
.
Universität Bielefeld
.
Ausblick
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Mögliche zukünftige Services
Linked Open Data
Vergabe von URIs für alle indexierten Dokumente
„BASE-Identifier“?
Identifier-Resolver
Ausgabe aller Identifier zu einer Publikation
Z.B.: arXiv:1105.0557 → doi:10.2140/pjm.2012.259.421
Repository Registry
Exponierung der Metadaten über Datenquellen
Das BASE-Netzwerk M. Lösch 31 / 35
.
Universität Bielefeld
.
Ausblick
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Mögliche zukünftige Services
Linked Open Data
Vergabe von URIs für alle indexierten Dokumente
„BASE-Identifier“?
Identifier-Resolver
Ausgabe aller Identifier zu einer Publikation
Z.B.: arXiv:1105.0557 → doi:10.2140/pjm.2012.259.421
Repository Registry
Exponierung der Metadaten über Datenquellen
Das BASE-Netzwerk M. Lösch 31 / 35
.
Universität Bielefeld
.
Ausblick
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Mögliche zukünftige Services
Linked Open Data
Vergabe von URIs für alle indexierten Dokumente
„BASE-Identifier“?
Identifier-Resolver
Ausgabe aller Identifier zu einer Publikation
Z.B.: arXiv:1105.0557 → doi:10.2140/pjm.2012.259.421
Repository Registry
Exponierung der Metadaten über Datenquellen
Das BASE-Netzwerk M. Lösch 31 / 35
.
Universität Bielefeld
.
Agenda
Voraussetzungen Services Ausblick Zusammenfassung Literatur
1. Voraussetzungen
2. Services
Suchschnittstelle
Rsync-Schnittstelle
OAI-Schnittstelle
Statistik-Schnittstelle
Klassifikationsschnittstelle
3. Ausblick
4. Zusammenfassung
Das BASE-Netzwerk M. Lösch 32 / 35
.
Universität Bielefeld
.
BASE-Netzwerk
Voraussetzungen Services Ausblick Zusammenfassung Literatur
.BASEAuthorClaim ViFas
OpenAIRE
Europeana
Cloud
Kataloge High
North
Das BASE-Netzwerk M. Lösch 33 / 35
.
Universität Bielefeld
Vielen Dank für die Aufmerksamkeit!
Das BASE-Netzwerk M. Lösch 34 / 35
.
Universität Bielefeld
.
Literatur
Voraussetzungen Services Ausblick Zusammenfassung Literatur
Kindling, M. und P. Schirmbacher (2013). Die digitale Forschungswelt
als Gegenstand der Forschung. Information – Wissenschaft &
Praxis 64(2-3), 127–136.
Vierkant, P., M. Voigt, J. Dupski, S. David und M. Lösch (2012). 2012
Census of Open Access Repositories. Poster presented at Open Access
Tage, Vienna (AT), 26.–27.9.2012.
Das BASE-Netzwerk M. Lösch 35 / 35


Das BASE-Netzwerk - Services und Applikationen im Umfeld der Bielefeld Academic Search Engine

Summann, Friedrich

Schirrwagen, Jochen

Dynamik in wissenschaftlicher Kommunikation und Publikation -> Funktion, Rolle und Strategien für Repositorien - Zukünftige Entwicklungen für Repositorien

Kienle, Andrea

Schlösser, Christian

Schlieker-Steens, Philipp

Prof. Dr. Andrea Kienle, Christian Schlösser, Philipp Schlieker-Steens 
University of Applied Sciences and Arts Dortmund, Emil-Figge-Str. 42, 44227 Dortmund 
(Germany), 
Email: andrea.kienle@fh-dortmund.de, christian.schloesser@fh-dortmund.de, philipp.schlieker-
steens@fh-dortmund.de 
Experiences from eye tracking chat-
communication using the INKA-SUITE 
Prof. Dr. Andrea Kienle, Christian Schlösser, Philipp Schlieker-Steens 
1 Introduction 
Eye tracking is a common way for analyzing the usability of applications (e.g., 
Duchowski, 2007), stationary single user systems (e.g., Shneiderman & Plaisant, 
2010) and websites (e.g., Nielsen & Loranger, 2006). Existing eye tracking sys-
tems are limited to static content like websites. In these applications the content is 
placed at the same position for all participants and can be easily compared. If this 
area of interest (AOI) is not static, but changes its size, shape or position, the eval-
uation becomes more complex and requires an amount of manual analysis. This is 
especially the case in cooperative applications like chats. The difficulty is, that 
you cannot predict the time of change, because it is caused by user input (e.g., new 
chat message, scrolling) - the dynamics of the AOI cannot be foretold. Existing 
eye tracking software is therefore not suitable in situations of changing AOIs. 
Therefore a new approach is needed to overcome these problems. 
The paper is structured by introducing you to the INKA-SUITE itself, followed by 
the description of a first experiment with its results. At the end there is an outlook. 
2 INKA-SUITE 
The INKA-SUITE is a platform that integrates the application and the analysis of 
eye tracking dynamic AOIs (e.g., Kienle et al., 2013). The base of the INKA-
SUITE is the connectivity to Tobii eye trackers (http://www.tobii.com). With a di-
rect connection between the eye tracker gaze data stream and GUI, the INKA-
SUITE identifies AOIs at runtime, regardless of size, shape and position. Every 
gaze data record is complemented by the underlying GUI element, identified by 
names, ID’s and/or references to other database tables such as chat messages or 
users. A subsequent manual work is eliminated and the evaluation can be started 
immediately. This profound cross-linking between analyses related functions and 
user software cannot be adapted by existing eye tracking analysis software. 
DOI: 10.2390/biecoll-saga2013_4
In Pfeiffer and Essig (2013): Proceedings of the First International Workshop on Solutions for Automatic Gaze-Data Analysis 2013 
(SAGA 2013). DOI: 10.2390/biecoll-saga2013_0
Experiences from eye tracking chat-communication using the INKA-SUITE 
The components of the INKA-SUITE are: 
(a) A server to control and manage clients as well as providing tem-
plates for the clients. Using templates enables researchers to conduct 
A/B tests. 
(b) A client application for user interaction, which is connected to an 
eye tracker and presents the chosen template. 
(c) An analysis component for managing projects and analyzing the 
collected data. Because fixations are represented by GUI elements, 
further information, such as fixation length and underlying AOI, can 
be retrieved through tooltips. 
Fig. 1: INKA-SUITE Timeline (left) and Replay (right) as part of the analysis component 
The analysis component processes the recorded data and is divided into three 
parts: 
(a) In the statistic tool, general information about the chat session and 
also about each user is presented. For example these are duration of 
chat, keystrokes per minute, time to first message and messages per 
minute. 
(b) Within the timeline tool, the entire chat session is presented on a 
timeline for each user (see Figure 1, left). This ensures the compara-
bility between users. Shortcuts, keystrokes, chat messages and fixa-
tions of the users are listed separately for each user. Using annota-
tions, important areas can be marked. The displayed output is fully 
dynamic and can be adapted to the current problem. The presentation 
of the collected data in form of the timeline, is oriented at the table 
structure of Beißwenger (2007) and the presentation of fixations by 
Stellmach et al. (2010). 
(c) The replay tool shows a replay of the chat from the user’s point of 
view (see Figure 1, right). This tool is chronology replaying the chat, 
similar to a screen recording, but with options to select and deselect 
the data that is output at runtime, also showing fixations and sac-
cades as a scanpath (e.g., Mealha et al., 2012). 
  Prof. Dr. Andrea Kienle, Christian Schlösser, Philipp Schlieker-Steens 
3 Experiment 
To test the INKA-SUITE, an experiment in cooperation with the Institute of Ger-
man language and literature of the TU Dortmund was accomplished. In this study 
the usability of the INKA-SUITE should be tested at first. Secondly questions 
from the linguists with respect to the influence of different user interfaces in chats 
for the structure of discussions should be answered.  
Fig. 2: Two templates: Standard Chat (left) and Talk (right) 
By using the templating feature two templates were created (see Figure 2). One 
template (Figure 2 left) represents a Standard-Chat-Environment (SCE) (e.g., 
Beißwenger, 2003) with an input field and a chat protocol. Template two (Figure 2 
right) is designed as a so called Talk layout similar to the Unix Talk. It contains 
the same features as the SCE but in addition a synchronous user list, which shows 
all connected chat partners and a live view from their text input fields. 
Tab.  1: Study organization 
The table above offers an overview about the experiment organization. In the 
study, 26 participants were recorded while chatting in two different scenarios. To 
solve the problems given by the scenarios, the participants have to exchange their 
information which differs to each other.  
The using of the INKA-SUITE while collecting and analyzing the data provides 
no problems. All ten chat-communications were completely recorded and supply a 
wide database for the analysis by the linguists. 
Day 1 - Chat Groups of two Groups of three 
Movie jury - scenario 2 1 
Murderer - scenario 2 
Day 2 - Talk Groups of two Groups of three 
Movie jury - scenario 2 1 
Murderer - scenario 2 
Experiences from eye tracking chat-communication using the INKA-SUITE 
Fig. 3: Study result – eye tracking inaccuracy 
The study revealed that the inaccuracy of the eye trackers is a problem (see Figure 
3). Some gaze plots do not land on the AOI which is fixated. Therefore a manual 
effort is needed in some cases. A solution could be repositioning and larger scal-
ing of the AOIs. The analysis by the linguists is still ongoing. 
4 Outlook 
The INKA-SUITE is basis for further research like automatic pattern recognition 
(e.g., revisions), contextualized communication and extension of simultaneous 
chats. A planned project called Chat++, enabled by real-time AOI identification, 
could therefore support features like reading awareness, eye tracking-based refer-
encing and activity and context-awareness. 
References 
1. Beißwenger, M. (2003): Sprachhandlungskoordination im Chat. In: Zeitschrift für german-
istische Linguistik 31 (2), S. 198–231.
2. Beißwenger, M. (2007): Sprachhandlungskoordination in der Chat-Kommunikation: Walter
de Gruyter (26).
3. Duchowski, Andrew (2007): Eye Tracking Methodology. Theory and Practice. Second.
London: Springer-Verlag London Limited. Online verfügbar unter
http://dx.doi.org/10.1007/978-1-84628-609-4.
4. Kienle, A.; Schlieker-Steens, P.; Schlösser C. (2013): INKA-SUITE: An Integrated Test-
Environment for Analyzing Chat Communication. In: CSCL 2013 - Conference Proceedings
Volume II, Conference Proceedings Volume II, S. 285–286.
5. Mealha, O.; Veloso, A.; Almeida, S.; Rodrigues, R.; Roque, L.; Marques, R.; Manteigueiro,
C. (2012): Eye tracking data representation and visualization: on information and communi-
cation studies at CETAC. MEDIA.
6. Nielsen, Jakob; Loranger, Hoa (2006): Prioritizing web usability: New Riders.
7. Shneiderman, Ben; Plaisant, Catherine (2010): Designing the user interface. Strategies for
effective human-computer interaction. 5. Aufl. Boston: Addison-Wesley.
8. Stellmach, Sophie; Nacke, Lennart; Dachselt, Raimund (2010): Advanced gaze visualiza-
tions for three-dimensional virtual environments. In: Proceedings of the 2010 symposium on
eye-tracking research & Applications. ACM, S. 109–112.


Experiences from eye tracking chat-communication using the INKA-SUITE

Peil, Vitali

PUB Linked Data Hub

Not a member yet