=Paper=
{{Paper
|id=Vol-1461/WOP2015_pattern_abstract_5
|storemode=property
|title=An Ontology Design Pattern for Particle Physics Analysis
|pdfUrl=https://ceur-ws.org/Vol-1461/WOP2015_pattern_abstract_5.pdf
|volume=Vol-1461
|dblpUrl=https://dblp.org/rec/conf/semweb/CarralCDHHHKLSV15
}}
==An Ontology Design Pattern for Particle Physics Analysis==
https://ceur-ws.org/Vol-1461/WOP2015_pattern_abstract_5.pdf
An Ontology Design Pattern for
Particle Physics Analysis.
David Carral,1 Michelle Cheatham,1 Sünje Dallmeier-Tiessen,2,3 Patricia
Herterich,2,4 Michael D. Hildreth,5 Pascal Hitzler,1 Adila Krisnadhi,1,8 Kati
Lassila-Perini,6 Elizabeth Sexton-Kennedy,2 Charles Vardeman,5 Gordon
Watts7
Wrigth State University,1 CERN,2 Harvard University,3 Humboldt-Universität zu
Berlin,4 University of Notre Dame,5 , Helsinki Institute of Physics6 ,
University of Washington,7 University of Indonesia8
Abstract. The detector final state is the core element of particle physics
analysis as it defines the physical characteristics that form the basis of
the measurement presented in a published paper. Although they are a
crucial part of the research process, detector final states are not yet
formally described, published in papers or searchable in a convenient
way. This paper aims at providing an ontology pattern for the detector
final state that can be used as a building block for an ontology covering
the whole particle physics analysis life cycle.
1 Introduction
Particle Physics, the study of the fundamental building blocks and forces of our
universe, involves some of the largest experimental apparatus ever constructed,
like the ALICE, ATLAS, CMS, and LHCb experiments located at the Large
Hadron Collider (LHC) at CERN. Each of these “experiments” is a very large
collaboration of physicists who work as a team to design, build, and operate
the particle detectors and to produce measurements characterizing the particles
that make up the universe. The measurements are inherently statistical in na-
ture: often billions or trillions of particle collisions are analyzed to determine
probabilities or probability densities associated with a given physical process.
Because many of the experiments collect multi-purpose data, careful attention
must be paid to defining the measurement that is to be made.
Despite the many thousands of papers published since the advent of particle
physics in the 1940s, the field has no formal way of representing or classifying
experimental results – no metadata accompanies an article to formally describe
the physics result therein. A number of scenarios would be enabled with such a
representation. For example, a physicist from ATLAS, or a theorist, could search
an external database for previous work done by CMS in order to compare results.
Even a physicist inside ATLAS could search an internal database for previous
examples similar to a planned analysis; a substantial amount of time and effort
can be saved by starting from some preexisting work.
We intend to address this situation with our ontology design pattern. Results
in particle physics take many forms, but all are based on the selection of a
target set of characteristics, a detector final state that defines the ingredients
of the measurement. The fundamental unit of particle physics is the individual
interaction of a set of particles, or an “event.” An event could, for example, be
captured from a single interaction of counter-rotating particles in a collider or
from the collision of a high-energy cosmic ray in the atmosphere. The selection
characteristics refer to properties of an event and can describe the presence or
absence of specific particles observed by the detector in the aftermath of the
collision, or potentially more global properties of the products produced in the
collision, such as the total energy released. Since the physics results we wish to
describe and preserve in a repository are all based on the selection of one or
more detector final states, this is a necessary ingredient of an ontology covering
the whole particle physics analysis life cycle.
Competency questions have been recognized as a good approach to detect and
generalize the modeling requirements from multiple domains that an ontology
can represent. They are queries that a domain expert would be expected to run
against a knowledge base. For the proposed final state ODP, such competency
questions include:
1. Retrieve all analyses requireing particles to have an invariant mass near the
Z pole.
2. Retrieve all analyses that used jets in the final state.
3. Retrieve all analyses that veto extra leptons.
4. Retrieve all analyses requiring large missing energy.
2 Formalization
This section presents the detector final state pattern by discussing the more
interesting classes, properties, and axioms. Description Logics (DL) [2] notation
is used to present the axioms. To encode the pattern, we make use of the logic
fragment SROIQ as defined in [4], which is the basis for the OWL 2 DL standard
[3]. The proposed ODP has been formally encoded using the Web Ontology
Language (OWL).1 A schematic view of the pattern is shown in Figure 1.
DetectorFinalState: A detector final state (DetectorFinalState) formally de-
scribes and structures information about a physics analysis (measurement) that
is defined by its use of a common set of particle physics characteristics. As such,
it must describe those characteristics of the fundamental “event” that have been
selected to make the measurement. It is defined, amongst other features, by a set
of particles/objects (PhysicsObjects) contained in the event and by global quan-
tities formed by performing some operation on the ensemble of objects contained
1
The pattern can be downloaded from
www.dropbox.com/sh/0upr45j1awd4q0d/AAAw9BQ2eZIWBIh_rBpP1Uu1a?dl=0.
Fig. 1. A schematic view of the DetectorFinalState ODP
in an event (EventLevelQuantity). Both particles/objects and the ensemble mea-
surements are referred to in this pattern as final state objects (FinalStateObject).
With the following axioms, we enforce that (1) every detector final state
must refer to at least final state object, (2) all final state objects are either event
level quantities or physics objects and (3) all event level quantities and physics
objects are final state objects.
DetectorFinalState v ∃referstTo.FinalStateObject (1)
FinalStateObject v EventLevelQuantity t PhysicsObject (2)
In order to select events of interest, these objects are subject to selection
criteria that are used to define a collection of events that serves as the basis for
a physics measurement, and hence must be captured in the pattern. A detector
final state (DetectorFinalState) then conveys numerical information describing
the selection. This numerical information is referred as the selection criteria
(SelectionCriteria) which models a complex boolean set of unary and binary re-
strictions. We make use of the classes And and Or to define complex selection
criteria.
DetectorFinalState v ∃hasSelectionCriteria.(SelectionCriteria t And t Or) (3)
And v ∃hasOperand.(SelectionCriteria t And t Or) (4)
Or v ∃hasOperand.(SelectionCriteria t And t Or) (5)
FinalStateObject: As mentioned above, there are two different types of final
state objects in our model: physics objects and event level quantities. Each
of these is defined by a restricted vocabulary and will point to another class,
namely BaseDefinition, which will serve as a hook to provide more specific infor-
mation about these types. Axiomatically, then, every PhysicsObject and every
EventLevelQuantity are FinalStateObjects:
PhysicsObject v FinalStateObject (6)
EventLevelQuantity v FinalStateObject (7)
In order for these quantities to have meaning, each of the FinalStateObjects
requires a BaseDefinition that describes the criteria for the creation of the Final-
StateObject.
An example typical selection could be “retrieve all detector final states in-
volving some electron with pT > 40 GeV.” In this case, the selection requires a
particular type of final state object and has a restriction of 40 GeV, a Physical-
Value with a NumericalValue of 40 and a Unit of GeV, on the PhysicalQuantity
pT , which is shorthand for the momentum of a particle in the plane transverse to
the beam axis. In general, a SelectionCriteria must indeed have a FirstArgument
specifying a value and a SecondArgument specifying of which PhysicalQuantity
this value is an instance, with some sort of binary operator (< or >, for example)
specifying the desired relationship.
For more information as to how information is stored using the pattern see
www.dropbox.com/sh/0upr45j1awd4q0d/AAAw9BQ2eZIWBIh_rBpP1Uu1a?dl=0. We
not only include terminological axioms in our ontology but also populate the
pattern using data from existing publications [1].
3 Conclusions and Future Work
This paper proposes a generic ODP to capture the common core of experimental
results from particle physics research. More specifically, it provides a precise
description of a detector final state which can be used to assign meaningful
metadata to the output produced by LHC. In future iterations we plan to extend
axiomatization and populate it using real-world data to validate its usability.
References
1. Aad, G., et al.: Search for pair-produced long-lived neutral particles decaying in
√
the ATLAS hadronic calorimeter in pp collisions at s = 8 TeV. Phys.Lett. B743,
15–34 (2015)
2. Baader, F., Calvanese, D., McGuinness, D., Nardi, D., Patel-Schneider, P. (eds.):
The Description Logic Handbook: Theory, Implementation, and Applications. Cam-
bridge University Press, second edn. (2007)
3. Hitzler, P., Krötzsch, M., Parsia, B., Patel-Schneider, P.F., Rudolph, S. (eds.):
OWL 2 Web Ontology Language: Primer. W3C Recommendation (27 October
2009), available at http://www.w3.org/TR/owl2-primer/
4. Horrocks, I., Kutz, O., Sattler, U.: The even more irresistible SROIQ. In: Proc.
of the 10th Int. Conf. on Principles of Knowledge Representation and Reasoning
(KR 2006). pp. 57–67. AAAI Press (2006)