=Paper=
{{Paper
|id=None
|storemode=property
|title=An Ontology-Based Tool for Collaborative and Social Sensemaking
|pdfUrl=https://ceur-ws.org/Vol-1051/paper5.pdf
|volume=Vol-1051
|dblpUrl=https://dblp.org/rec/conf/ihc/GarciaPP13
}}
==An Ontology-Based Tool for Collaborative and Social Sensemaking==
https://ceur-ws.org/Vol-1051/paper5.pdf
An Ontology-Based Tool for Collaborative and Social
Sensemaking
Ana Cristina Bicharra Fernando Pinto Nayat Sanchez-Pi
ADDLabs. ADDLabs. ADDLabs.
Computer Science Computer Science Computer Science
Department. Fluminense Department. Fluminense Department. Fluminense
Federal University Federal University Federal University
Rua Passo da Pátria, 156 Rua Passo da Pátria, 156 Rua Passo da Pátria, 156
Niterói. RJ,Brazil Niterói. RJ,Brazil Niterói. RJ,Brazil
bicharra@ic.uff.br fernando@addlabs.uff.br nayat@addlabs.uff.br
ABSTRACT helps users in focussing their attention and comprehension
Sensemaking activities of social networks involve network of root causes of the problem. Further work is needed to
exploration and representation so, visual tools are designed develop more fully intuitive visualizations that exploit the
to support these two activities. Existing social network anal- richer information and make the multiple connections be-
ysis tools are usually weak in supporting complex analytical tween data more easily accessible.
tasks and also in providing a collaborative environment for
interaction. The analysis of data using a visual tool is rarely Categories and Subject Descriptors
a task done in isolation, it tends to be part of a wider goal:
that of making sense of the current situation, often to sup- H.5.2 [Information Interfaces and Presentation]: User
port decision-making. This paper discusses the storytelling Interfaces; D.2.2 [Software Engineering]: Design Tools
design of a software environment to support organizations and Techniques; L.1.3 [Knowledge and Media]: Ontol-
in sense-making activities and to support accidents investi- ogy/Taxonomy and Classification
gation. A case study ACR-C describes petroleum industry
employees investigating the root cause of an accident issue General Terms
observed in one (or more) platforms. It is used through-
out the paper as an example of human computer interaction HCI
where the ontology becomes a tool with domain knowledge
to assist expert persons building a root cause tree leading Keywords
to accidents. The framework will also provide with a col-
laborative recommendation module assuming that the users collaborative analysis recommendation, social network, root
build up clusters based on their similar analysis in rating of cause analysis, ontology
items. A case study ACR-C describes petroleum industry
employees investigating the root cause of an accident issue 1. INTRODUCTION
observed in one (or more) platforms. It is used through-
out the paper as an example of human computer interaction There is an important effort of oil and gas industry to
where the ontology becomes a tool with domain knowledge reduce the number of accidents and incidents. There are
to assist expert persons buildind a root cause tree leading standards to identify and record workplace accidents and
to accidents. This paper reports the experience gained in incidents to provide guiding means on prevention efforts, in-
ACR-C, a project that aims to support knowledge manage- dicating specific failures or reference, means of correction
ment (KM), sharing and reuse across different media in oil of conditions or circumstances that culminated in accident.
& gas industry. We report the storytelling design approach Besides, oil and gas industry is increasingly concerned with
adopted and the design phases that led to the first prototype. achieving and demonstrating good performance of occupa-
A user interface was designed to assess how different levels of tional health and safety (OHS), through the control of its
data, information and knowledge were mapped using alter- risks, consistent with its policy and objectives.
native visual tools. The results show that a clear separation Even if the focus on risk management is increasing in our
of the visual data analysis from other sense-making subtasks society, major accidents resulting in several fatalities seem to
be unavoidable in some industries. Since the consequences of
such major accidents are unacceptable, a thorough investi-
gation of the accidents should be performed in order to learn
from what has happened, and prevent future accidents.
Today, with the advances of new technologies, accidents,
incidents and occupational health records are stored in het-
Copyright by the paper’s authors. Copying permitted only for private and erogeneous repositories. During the last decades, a number
academic purposes. In: Proceedings of the V Workshop sobre Aspectos da of methods for accident investigation have been developed.
Interação Humano-Computador na Web Social (WAIHCWS’13), Manaus,
Brazil, 2013, published at http://ceur-ws.org. Each of these methods has different areas of application and
. different qualities and deficiencies. This poses a top priority
34
challenge for oil & gas industries that are looking for inno- • Safety-based RCA descends from the fields of accident
vative ways to design human-computer interaction, in order analysis and occupational safety and health.
to extract knowledge from masses of data.
Besides with the recent advances in technology, there is • Production-based RCA has its origins in the field of
an emerging presence of social media and social networking quality control for industrial manufacturing.
systems. The reason is that despite there is an increasing • Process-based RCA is basically a follow-on to production-
interest in the exploration of social networks, there does not based RCA, but with a scope that has been expanded
exist a concrete dataset that includes both explicit bonds of to include business processes.
personalized characteristics among users and a collaborative
annotation of items. This is due to that most social media • Failure-based RCA is rooted in the practice of failure
systems do not allow for free access to all user profiles or analysis as employed in engineering and maintenance.
lists of friends.
Moreover, the need for computer supported collaboration • Systems-based RCA has emerged as an amalgamation
has grown over the last years and made collaboration pro- of the preceding schools, along with ideas taken from
cesses an important factor within organizations [1]. This fields such as change management, risk management,
trend has resulted in the development of a variety of tools and systems analysis.
and technologies to support the various forms of collabora-
tion. 3. PROPOSAL: ACR-C
Managing conflicts in human-computer interaction poses a The ACR system aims to assist in the investigation by
set of challenges beyond those encountered in dealing strictly identifying the root causes of anomalies, and lead the user
with software. Some familiar issues arise such as asymme- to create, query and visualize trees root cause, considering
try in capabilities and responsibilities distributed processing the need for data entry during the process. The system
and information storage limited resources and a high cost of guides the user in the investigation of the root cause of an
exchanging information. Some of the works in the HCI field anomaly, through questions and the construction of a fault
are [10], [11], [8],[12] tree, an interactive walkthrough. It permits the creation of
Confirming the above, there exist several research projects a collaborative environment of research, using an ontology
in this area and MIT Deliberatorium Project from the MIT on accident investigation as a thread of communication and
Center for Collective Intelligence is one of them, i.e [6, 7], cooperation among the actors who participate in the inves-
arguing the importance of the aggregated value of collabo- tigative process group.
ration process. The ontology permits the use of a common language for
Given the incentives of the widespread adoption of social the actors in the investigative process group. ACR tool al-
networks for collaborative recommendations and of the lack low not only to investigate the root cause of accidents but
of some previous study that directly addresses the problem also to have an aggregated register to be used for future or
of efficiently integrating the added- value knowledge pro- past analysis. It also provide with an interaction blackboard
vided by those networks in the field of collaborative recom- where user communication takes place. To accomplish with
mendation, we propose a the storytelling design of a collab- the hypothesis presented we built an explicit domain on-
orative environment [2] to support oil and gas industry in tology in order to allow a common communication among
sense-making activities. We extend a similar work proposed participants. Sometimes in the early commitment decision
in literaure [3]. Our case study ACR-C describes petroleum process occur an early arrival to conclusions or the failing to
industry employees investigating the root cause of an acci- share views on research, so a domain ontology allows facts
dent issue observed in one (or more) platforms. It is used coexist avoiding conflicting information. It also make clearer
throughout the paper as an example of human computer in- the information underlying the conclusions of each of the
teraction where the ontology becomes a tool with domain participants, and it does not impose coordination of actions
knowledge to assist expert persons building a root cause between participants.
tree leading to accidents. The framework will also provide
with a collaborative recommendation module assuming that 3.1 ACR-C: Domain Model
the users build up clusters based on their similar analysis The ontology system ACR contains classes, attributes, do-
in rating of items. A model will learn based on patterns main values, causes, facts and assumptions to create the con-
recognized in the rating analysis of users using clustering, cepts and constraints of the system, and also their knowledge
Bayesian networks and other machine learning techniques structure. See 1.
will be applied. Concepts in the ontology are used to define objects collec-
tions with similar characteristics. In the case of ACR, main
2. ROOT CAUSE ANALYSIS concepts are identified as:
The Root Cause Analysis is a methodology that proves • Fault Tree: It is the high level concept representing
to be essential for any organization, especially for indus- the domain on discourse. It is composed of causes, hy-
trial operations that needs to eliminate the recurrence of potheses, evidence and observations of failures (facts)
failures and accidents. [5]. Root cause analysis is not a related to the occurrence of an unwanted event, called
single, sharply defined methodology; there are many differ- anomaly.
ent tools, processes, and philosophies for performing RCA
[9]. However, several very-broadly defined approaches can • Anomaly: Description of an undesirable event or sit-
be identified by their basic approach or field of origin: safety- uation which results or may result in damage or failure,
based, production-based, process-based, failure-based, and affecting people, the environment, equity (own or third
systems-based. party), products or processes.
35
Figure 1: ACR-C ontology
– Deviation: Any action or condition that has the or upon satisfaction of other physiological needs
potential to lead to, directly or indirectly, damage at the workplace or during this, the employee is
to people, to property (own or third party) or considered in carrying out the work.
environmental impact, which is inconsistent with
labor standards, procedures, legal or regulatory • Fact: It’s some event about which there is no doubt.
requirements, requirements management system It is an event that has been observed by the research
or practice. team and, at first, was direct cause of the anomaly.
∗ Behavioral deviation Act or omission which, During investigation, however, it is possible for the
contrary provision of security, may cause or user to reclassify a fact turning it into developing a
contribute to the occurrence of accidents. hypothesis.
∗ Non-behavioral deviation: Environmen-
• Hypothesis: It is an assumption that makes the oc-
tal condition that can cause an accident or
currence of an event that may have contributed to the
contribute to its occurrence. The environ-
occurrence of the anomaly.
ment includes adjective here, everything that
relates to the environment, from the atmo-
– Confirmed Hypothesis: Kind of hypothesis whose
sphere of the workplace to the facilities, equip-
evidence, explicit in the model or only in the mind
ment, materials used and methods of work-
of the user, leading the user to decide to confirm
ing employees who is inconsistent with labor
it turning it into a cause or root cause.
standards, procedures, legal requirements or
normative requirements of the management – Hypothesis in development: Hypothesis type
system or practice. which is still being investigated.
– Incident: Any evidence, personal occurrence or – Hypothesis cancelled: Kind of hypothesis whose
condition that relates to the environment and/or evidence, explicit in the model or only in the mind
working conditions, can lead to damage to physi- of the user, take the user to decide to discard it,
cal and/or mental. transforming it into a hypothesis denied and end-
– Accident: Occurrence of unexpected and unwel- ing a line of investigation.
come, instant or otherwise, related to the exercise
of the job, which results or may result in personal The rest of the concepts and its relations can be seen in
injury. The accident includes both events that Figure 1. It is considered as a basis for the use of root cause
may be identified in relation to a particular time analysis in investigative processes or hypothesis. With the
or occurrences as continuous or intermittent ex- export interface, the customer can add to the information
posure, which can only be identified in terms of that has already been registered in the tree, as well as re-
time period probable. A personal injury includes quest an export of a specific tree.
both traumatic injuries and illnesses, as damaging
effects mental, neurological or systemic, resulting 3.2 ACR-C: Interaction Model
from exposures or circumstances prevailing at the They are computer systems designed to enhance the per-
year’s work force. In the period for meal or rest, formance of work in groups. These computational tools
36
Figure 2: ACR interface.
Figure 3: ACR interface. Entering evidences
37
should be modeled in order to foster interaction among par- Encyclopedia of E-Collaboration, Ned Kock (org),
ticipants, serving as a facilitator for coordination, collabora- pages 637–644, 2007.
tion and communication between participants that make up [5] E. B. Jensen. Root cause analysis. Compendium for
the group, both in the same location as at spatially different use by Patient, 2004.
locations [4]. [6] M. Klein. How to harvest collective wisdom on
It takes the user to describe the anomaly (accident / in- complex problems: An introduction to the mit
cident / deviation), create chances and give evidence, to deliberatorium. Center for Collective Intelligence
confirm or rule out a hypothesis. Evidence given a chance working paper, 2011.
by the user, during the investigation, can serve as input for [7] M. Klein. The mit deliberatorium: Enabling
the user himself decides to quit, or follow, in particular line large-scale deliberation about complex systemic
of research. The development of hypotheses lead to the root problems. In Collaboration Technologies and Systems
causes of the problem. (CTS), 2011 International Conference on, pages
This aggregated value provides an environment where tech- 161–161. IEEE, 2011.
nical specialists could collaboratively solve problems and [8] P. Lévy and R. Bonomo. Collective intelligence:
identify and share best practices. This tool was modeled Mankind’s emerging world in cyberspace. Perseus
for accomplish an easy and comprehensive user interaction. Publishing, 1999.
The main characteristic is that the system acts as a black-
[9] W. Runciman, P. Hibbert, R. Thomson, T. Van
board where all the communication process between actors
Der Schaaf, H. Sherman, and P. Lewalle. Towards an
take place with the information organized in fact- hypothe-
international classification for patient safety: key
sis method because actors have an a common environment
concepts and terms. International Journal for Quality
to collaborate, synchronous or asynchronous. See 2 and 3.
in Health Care, 21(1):18–26, 2009.
[10] R. Shaw and M. Davis. Toward emergent
4. CONCLUSIONS representations for video. In Proceedings of the 13th
This paper describes a preliminary work in the creation annual ACM international conference on Multimedia,
of a collaborative environment for accident investigation, us- pages 431–434. ACM, 2005.
ing an ontology-based system as a thread of communication [11] R. Shaw and P. Schmitz. Community annotation and
and cooperation among the actors who participate in the remix: a research platform and pilot deployment. In
investigative process group. Proceedings of the 1st ACM international workshop on
We built an explicit domain ontology in order to allow a Human-centered multimedia, pages 89–98. ACM, 2006.
common communication among participants. It also allows [12] D. Zhang, B. Guo, and Z. Yu. The emergence of social
facts coexist avoiding conflicting information. It also makes and community intelligence. Computer, 44(7):21–28,
clearer the information underlying the conclusions of each 2011.
of the participants, and it does not impose coordination of
actions between participants.
A case study was presented to illustrate the functionalities
of the system and also how the interaction presents different
colors for entities linked to the same hypothesis turning very
easy the comprehension.
As future work we are still working on the collaborative
recommendation module that will assume users build up
clusters based on their similar analysis in rating of items.
A model will learn based on patterns recognized in the rat-
ing analysis of users using clustering, Bayesian networks and
other machine learning techniques will be applied.
5. REFERENCES
[1] D. Azevedo, J. Janeiro, S. Lukosch, R. O. Briggsc, and
B. Fonsecaa. An integrative approach to
diagram-based collaborative brainstorming. In
Proceedings of the ECSCW 2011 Workshop on
Collaborative usage and development of models and
visualizations, 2011.
[2] S. K. Card, T. P. Moran, and A. Newell. The
keystroke-level model for user performance time with
interactive systems. Communications of the ACM,
23(7):396–410, 1980.
[3] K. Chorianopoulos. Collective intelligence within web
video. Human-centric Computing and Information
Sciences, 3(1):10, 2013.
[4] H. Fuks, A. Raposo, M. A. Gerosa, M. Pimentel, and
C. J. Lucena. The 3c collaboration model. The
38