2015
Partially observable Markov decision processes (POMDPs) and the belief-desire-intention (BDI) framework have several complimentary strengths. We propose an agent architecture which combines these two powerful approaches to capitalize on their strengths. Our architecture introduces the notion of intensity of the desire for a goal’s achievement. We also define an update rule for goals’ desire levels. When to select a new goal to focus on is also defined. To verify that the proposed architecture works, experiments were run with an agent based on the architecture, in a domain where multiple goals must continually be achieved. The results show that (i) while the agent is pursuing goals, it can concurrently perform rewarding actions not directly related to its goals, (ii) the trade-off between goals and preferences can be set effectively and (iii) goals and preferences can be satisfied even while dealing with stochastic actions and perceptions. We believe that the proposed architecture furthers the theory of high-level autonomous agent reasoning.
@{103,
author = {Gavin Rens and Tommie Meyer},
title = {Hybrid POMDP-BDI: An Agent Architecture with Online Stochastic Planning and Desires with Changing Intensity Levels},
abstract = {Partially observable Markov decision processes (POMDPs) and the belief-desire-intention (BDI) framework have several complimentary strengths. We propose an agent architecture which combines these two powerful approaches to capitalize on their strengths. Our architecture introduces the notion of intensity of the desire for a goal’s achievement. We also define an update rule for goals’ desire levels. When to select a new goal to focus on is also defined. To verify that the proposed architecture works, experiments were run with an agent based on the architecture, in a domain where multiple goals must continually be achieved. The results show that (i) while the agent is pursuing goals, it can concurrently perform rewarding actions not directly related to its goals, (ii) the trade-off between goals and preferences can be set effectively and (iii) goals and preferences can be satisfied even while dealing with stochastic actions and perceptions. We believe that the proposed architecture furthers the theory of high-level autonomous agent reasoning.},
year = {2015},
journal = {International Conference on Agents and Artificial Intelligence (ICAART) Vol. 1},
pages = {5-14},
month = {10/01-12/01},
isbn = {978-989-758-073-4},
}
We present a decidable logic in which queries can be posed about (i) the degree of belief in a propositional sentence after an arbitrary finite number of actions and observations and (ii) the utility of a finite sequence of actions after a number of actions and observations. Another contribution of this work is that a POMDP model specification is allowed to be partial or incomplete with no restriction on the lack of information specified for the model. The model may even contain information about non-initial beliefs. Essentially, entailment of arbitrary queries (expressible in the language) can be answered. A sound, complete and terminating decision procedure is provided.
@{102,
author = {Gavin Rens and Tommie Meyer and G. Lakemeyer},
title = {A Modal Logic for the Decision-Theoretic Projection Problem},
abstract = {We present a decidable logic in which queries can be posed about (i) the degree of belief in a propositional sentence after an arbitrary finite number of actions and observations and (ii) the utility of a finite sequence of actions after a number of actions and observations. Another contribution of this work is that a POMDP model specification is allowed to be partial or incomplete with no restriction on the lack of information specified for the model. The model may even contain information about non-initial beliefs. Essentially, entailment of arbitrary queries (expressible in the language) can be answered. A sound, complete and terminating decision procedure is provided.},
year = {2015},
journal = {International Conference on Agents and Artificial Intelligence (ICAART) Vol. 2},
pages = {5-16},
month = {10/01-12/01},
isbn = {978-989-758-074-1},
}
2014
Poor adherence to prescribed treatment is a major factor contributing to tuberculosis patients developing drug resistance and failing treatment. Treatment adherence behaviour is influenced by diverse personal, cultural and socio-economic factors that vary between regions and communities. Decision network models can potentially be used to predict treatment adherence behaviour. However, determining the network structure (identifying the factors and their causal relations) and the conditional probabilities is a challenging task. To resolve the former we developed an ontology supported by current scientific literature to categorise and clarify the similarity and granularity of factors
@{158,
author = {Olukunle Ogundele and Deshen Moodley and Chris Seebregts and Anban Pillay},
title = {Building Semantic Causal Models to Predict Treatment Adherence for Tuberculosis Patients in Sub-Saharan Africa},
abstract = {Poor adherence to prescribed treatment is a major factor contributing to tuberculosis patients developing drug resistance and failing treatment. Treatment adherence behaviour is influenced by diverse personal, cultural and socio-economic factors that vary between regions and communities. Decision network models can potentially be used to predict treatment adherence behaviour. However, determining the network structure (identifying the factors and their causal relations) and the conditional probabilities is a challenging task. To resolve the former we developed an ontology supported by current scientific literature to categorise and clarify the similarity and granularity of factors},
year = {2014},
journal = {4th International Symposium (FHIES 2014) and 6th International Workshop (SEHC 2014)},
pages = {81-95},
month = {17/07-18/07},
}
Formalized temporal attributes in temporal conceptual models, temporal EER model, using a temporal description logic language (DLRus). This ensured the full formalization of the temporal conceptual model, ERvt, which permits full reasoning on the model. These results permit the development of consistent temporal databases.
@phdthesis{112,
author = {Nasubo Ongoma},
title = {Formalising Temporal Attributes in Temporal Conceptual Data Models},
abstract = {Formalized temporal attributes in temporal conceptual models, temporal EER model, using a temporal description logic language (DLRus). This ensured the full formalization of the temporal conceptual model, ERvt, which permits full reasoning on the model. These results permit the development of consistent temporal databases.},
year = {2014},
volume = {MSc},
}
One of the challenges in the Software Development Life Cycle (SDLC) is to ensure that the requirements that drive the development of a software system are correct. However, establishing unambiguous and error-free requirements is not a trivial problem. As part of the requirements phase of the SDLC, a conceptual model can be created which describes the objects, relationships and operations that are of importance to business. Such a conceptual model is often expressed as a UML class diagram. Recent research concerned with the formal validation of such UML class diagrams has focused on transforming UML class diagrams to various formalisms such as description logics. Description logics are desirable since they have reasoning support which can be used to show that a UML class diagram is consistent/inconsistent. Yet, even when a UML class diagram is consistent, it still does not address the problem of ensuring that a UML class diagram represents business requirements accurately. To validate such diagrams business analysts use a technique called scenario testing. In this paper we present an approach for the formal validation of UML class diagrams based on scenario testing. We additionally provide preliminary feedback on the experiences gained from using our scenario testing approach on a real-world software project.
@misc{106,
author = {Henriette Harmse and Katarina Britz and Aurona Gerber and Deshen Moodley},
title = {Scenario Testing using Formal Ontologies},
abstract = {One of the challenges in the Software Development Life Cycle (SDLC) is to ensure that the requirements that drive the development of a software system are correct. However, establishing unambiguous and error-free requirements is not a trivial problem. As part of the requirements phase of the SDLC, a conceptual model can be created which describes the objects, relationships and operations that are of importance to business. Such a conceptual model is often expressed as a UML class diagram. Recent research concerned with the formal validation of such UML class diagrams has focused on transforming UML class diagrams to various formalisms such as description logics. Description logics are desirable since they have reasoning support which can be used to show that a UML class diagram is consistent/inconsistent. Yet, even when a UML class diagram is consistent, it still does not address the problem of ensuring that a UML class diagram represents business requirements accurately. To validate such diagrams business analysts use a technique called scenario testing. In this paper we present an approach for the formal validation of UML class diagrams based on scenario testing. We additionally provide preliminary feedback on the experiences gained from using our scenario testing approach on a real-world software project.},
year = {2014},
isbn = {urn:nbn:de:0074-1301-3},
url = {http://ceur-ws.org/Vol-1301/ontocomodise2014_10.pdf},
}
The goal of this paper is to report on the development of an ontologybased taxonomic key application that is a first deliverable of a larger project that has as goal the development of ontology-driven computing solutions for problems experienced in taxonomy. The ontology-based taxonomic key was developed from a complex taxonomic data set, namely the Catalogue of Afrotropical Bees. The key is used to identify the genera of African bees and for this paper we developed an ontology-based application, that demonstrates that morphological key data can be captured effectively in a standardised format as an ontology, and furthermore, even though the ontology-based key provides the same identification results as the traditional key, this approach allows for several additional advantages that could support taxonomy in the biological sciences. The morphology ontology for Afrotropical bees, as well as the key application form the basis of a suite of tools that we intend to develop to support the taxonomic processes in this domain.
@inbook{101,
author = {Aurona Gerber and C. Eardley and Nishal Morar},
title = {An Ontology-based Key for Afrotropical Bees},
abstract = {The goal of this paper is to report on the development of an ontologybased
taxonomic key application that is a first deliverable of a larger project that
has as goal the development of ontology-driven computing solutions for problems
experienced in taxonomy. The ontology-based taxonomic key was developed from
a complex taxonomic data set, namely the Catalogue of Afrotropical Bees. The
key is used to identify the genera of African bees and for this paper we developed
an ontology-based application, that demonstrates that morphological key data can
be captured effectively in a standardised format as an ontology, and furthermore,
even though the ontology-based key provides the same identification results as the
traditional key, this approach allows for several additional advantages that could
support taxonomy in the biological sciences. The morphology ontology for
Afrotropical bees, as well as the key application form the basis of a suite of tools
that we intend to develop to support the taxonomic processes in this domain.},
year = {2014},
journal = {Frontiers in Artificial Intelligence and Applications},
pages = {277-288},
publisher = {IOS Press},
address = {Rio de Janeiro, Brazil},
isbn = {978-1-61499-437-4 (print) | 978-1-61499-438-1 (online)},
url = {http://ebooks.iospress.nl/volume/formal-ontology-in-information-systems-proceedings-of-the-eighth-international-conference-fois-2014},
}
The thesis reports on the development of a sequence of logics (formal languages based on mathematical logic) to deal with a class of uncertainty that agents may encounter. More accurately, the logics are meant to be used for allowing robots or software agents to reason about the uncertainty they have about the effects of their actions and the noisiness of their observations. The approach is to take the well-established formalism called the partially observable Markov decision process (POMDP) as an underlying formalism and then design a modal logic based on POMDP theory to allow an agent to reason with a knowledge-base (including knowledge about the uncertainties). First, three logics are designed, each one adding one or more important features for reasoning in the class of domains of interest (i.e., domains where stochastic action and sensing are considered). The final logic, called the Stochastic Decision Logic (SDL) combines the three logics into a coherent formalism, adding three important notions for reasoning about stochastic decision-theoretic domains: (i) representation of and reasoning about degrees of belief in a statement, given stochastic knowledge, (ii) representation of and reasoning about the expected future rewards of a sequence of actions and (iii) the progression or update of an agent’s epistemic, stochastic knowledge. For all the logics developed in this thesis, entailment is defined, that is, whether a sentence logically follows from a knowledge-base. Decision procedures for determining entailment are developed, and they are all proved sound, complete and terminating. The decision procedures all employ tableau calculi to deal with the traditional logical aspects, and systems of equations and inequalities to deal with the probabilistic aspects. Besides promoting the compact representation of POMDP models, and the power that logic brings to the automation of reasoning, the Stochastic Decision Logic is novel and significant in that it allows the agent to determine whether or not a set of sentences is entailed by an arbitrarily precise specification of a POMDP model, where this is not possible with standard POMDPs. The research conducted for this thesis has resulted in several publications and has been presented at several workshops, symposia and conferences.
@phdthesis{100,
author = {Gavin Rens},
title = {Formalisms for Agents Reasoning with Stochastic Actions and Perceptions},
abstract = {The thesis reports on the development of a sequence of logics (formal languages based on mathematical logic) to deal with a class of uncertainty that agents may encounter. More accurately, the logics are meant to be used for allowing robots or software agents to reason about the uncertainty they have about the effects of their actions and the noisiness of their observations. The approach is to take the well-established formalism called the
partially observable Markov decision process (POMDP) as an underlying formalism and then design a modal logic based on POMDP theory to allow an agent to reason with a knowledge-base (including knowledge about the uncertainties).
First, three logics are designed, each one adding one or more important features for reasoning in the class of domains of interest (i.e., domains where stochastic action and sensing are considered). The final logic, called the Stochastic Decision Logic (SDL) combines the three logics into a coherent formalism, adding three important notions for reasoning about stochastic decision-theoretic domains: (i) representation of and reasoning about degrees of belief in a statement, given stochastic knowledge, (ii) representation of and reasoning about the expected future rewards of a sequence of actions and (iii) the progression or update of an agent’s epistemic, stochastic knowledge.
For all the logics developed in this thesis, entailment is defined, that is, whether a sentence logically follows from a knowledge-base. Decision procedures for determining entailment are developed, and they are all proved sound, complete and terminating. The decision procedures all employ tableau calculi to deal with the traditional logical aspects, and systems of equations and inequalities to deal with the probabilistic aspects.
Besides promoting the compact representation of POMDP models, and the power that logic brings to the automation of reasoning, the Stochastic Decision Logic is novel and significant in that it allows the agent to determine whether or not a set of sentences is entailed by an arbitrarily precise specification of a POMDP model, where this is not possible with standard POMDPs.
The research conducted for this thesis has resulted in several publications and has been presented at several workshops, symposia and conferences.},
year = {2014},
volume = {PhD},
}
No Abstract
@misc{96,
author = {Nasubo Ongoma},
title = {Formalising Temporal Attributes in Temporal conceptual data models},
abstract = {No Abstract},
year = {2014},
}
The sugarcane supply chain (from sugarcane grower to mill) have particular challenges. One of these is that the growers have to deliver their cane to the mill before its quality degrades. The sugarcane supply chain typically consists of many growers and a mill. Growers deliver their cane daily during the milling season; the amount of cane they deliver depends on their farm size. Growers make decisions about when to harvest the cane, and the number and type of trucks needed to deliver their cane. The mill wants a consistent cane supply over the milling season. Growers are sometimes affected long queue lengths at the mill when they offload their cane. A preliminary agent-based simulation model was developed to understand this complex system. The model inputs a number of growers, and the amount of cane they are to deliver over the milling season. The number of trucks needed by each grower is determined by the trip, loading and unloading times and the anticipated waiting time at the mill. The anticipated waiting time was varied to determine how many trucks would be needed in the system to deliver the week’s cane allocation. As the anticipated waiting time increased, the number of trucks needed also increased, which in turn delayed the trucks when queuing at the mill. The growers’ anticipated waiting times never matched the actual waiting times. The research shows the promise of agent-based models as a sense-making approach to understanding systems where there are many individuals who have autonomous behaviour, and whose actions and interactions can result in unexpected system-level behaviour.
@{94,
author = {C. Sue Price and Deshen Moodley and C.N. Bezuidenhout},
title = {Using agent-based simulation to explore sugarcane supply chain transport complexities at a mill scale},
abstract = {The sugarcane supply chain (from sugarcane grower to mill) have particular challenges. One of these is that the growers have to deliver their cane to the mill before its quality degrades. The sugarcane supply chain typically consists of many growers and a mill. Growers deliver their cane daily during the milling season; the amount of cane they deliver depends on their farm size. Growers make decisions about when to harvest the cane, and the number and type of trucks needed to deliver their cane. The mill wants a consistent cane supply over the milling season. Growers are sometimes affected long queue lengths at the mill when they offload their cane.
A preliminary agent-based simulation model was developed to understand this complex system. The model inputs a number of growers, and the amount of cane they are to deliver over the milling season. The number of trucks needed by each grower is determined by the trip, loading and unloading times and the anticipated waiting time at the mill. The anticipated waiting time was varied to determine how many trucks would be needed in the system to deliver the week’s cane allocation. As the anticipated waiting time increased, the number of trucks needed also increased, which in turn delayed the trucks when queuing at the mill. The growers’ anticipated waiting times never matched the actual waiting times. The research shows the promise of agent-based models as a sense-making approach to understanding systems where there are many individuals who have autonomous behaviour, and whose actions and interactions can result in unexpected system-level behaviour.},
year = {2014},
journal = {43rd Annual Operations Research Society of South Africa Conference},
pages = {88-96},
month = {14/09-17/09},
isbn = {978-1-86822-656-6},
}
South Africa has one of the highest HIV infection rates in the world with more than 5.6 million infected people and consequently has the largest antiretroviral treatment program with more than 1.5 million people on treatment. The development of drug resistance is a major factor impeding the efficacy of antiretroviral treatment. While genotype resistance testing (GRT) is the standard method to determine resistance, access to these tests is limited in resource-limited settings. This research investigates the efficacy of multi-label machine learning techniques at predicting HIV drug resistance from routine treatment and laboratory data. Six techniques, namely, binary relevance, HOMER, MLkNN, predictive clustering trees (PCT), RAkEL and ensemble of classifier chains (ECC) have been tested and evaluated on data from medical records of patients enrolled in an HIV treatment failure clinic in rural KwaZulu-Natal in South Africa. The performance is measured using five scalar evaluation measures and receiver operating characteristic (ROC) curves. The techniques were found to provide useful predictive information in most cases. The PCT and ECC techniques perform best and have true positive prediction rates of 97% and 98% respectively for specific drugs. The ECC method also achieved an AUC value of 0.83, which is comparable to the current state of the art. All models have been validated using 10 fold cross validation and show increased performance when additional data is added. In order to make use of these techniques in the field, a tool is presented that may, with small modifications, be integrated into public HIV treatment programs in South Africa and could assist clinicians to identify patients with a high probability of drug resistance.
@phdthesis{93,
author = {Pascal Brandt and Deshen Moodley and Anban Pillay},
title = {An Investigation Of Multi-label Classification Techniques For Predicting Hiv Drug Resistance In Resource-limited Settings},
abstract = {South Africa has one of the highest HIV infection rates in the world with more than 5.6 million infected people and consequently has the largest antiretroviral treatment program with more than 1.5 million people on treatment. The development of drug resistance is a major factor impeding the efficacy of antiretroviral treatment. While genotype resistance testing (GRT) is the standard method to determine resistance, access to these tests is limited in resource-limited settings. This research investigates the efficacy of multi-label machine learning techniques at predicting HIV drug resistance from routine treatment and laboratory data. Six techniques, namely, binary relevance, HOMER, MLkNN, predictive clustering trees (PCT), RAkEL and ensemble of classifier chains (ECC) have been tested and evaluated on data from medical records of patients enrolled in an HIV treatment failure clinic in rural KwaZulu-Natal in South Africa. The performance is measured using five scalar evaluation measures and receiver operating characteristic (ROC) curves. The techniques were found to provide useful predictive information in most cases. The PCT and ECC techniques perform best and have true positive prediction rates of 97% and 98% respectively for specific drugs. The ECC method also achieved an AUC value of 0.83, which is comparable to the current state of the art. All models have been validated using 10 fold cross validation and show increased performance when additional data is added. In order to make use of these techniques in the field, a tool is presented that may, with small modifications, be integrated into public HIV treatment programs in South Africa and could assist clinicians to identify patients with a high probability of drug resistance.},
year = {2014},
volume = {MSc},
}
eHealth governance and regulation are necessary in low resource African countries to ensure effective and equitable use of health information technology and to realize national eHealth goals such as interoperability, adoption of standards and data integration. eHealth regulatory frameworks are under-developed in low resource settings, which hampers the progression towards coherent and effective national health information systems. Ontologies have the potential to clarify issues around interoperability and the effectiveness of different standards to deal with different aspects of interoperability. Ontologies can facilitate drafting, reusing, implementing and compliance testing of eHealth regulations. In this regard, we have developed an OWL ontology to capture key concepts and relations concerning interoperability and standards. The ontology includes an operational definition for interoperability and is an initial step towards the development of a knowledge representation modeling platform for eHealth regulation and governance.
@{92,
author = {Deshen Moodley and Chris Seebregts and Anban Pillay and Tommie Meyer},
title = {An Ontology for Regulating eHealth Interoperability in Developing African Countries},
abstract = {eHealth governance and regulation are necessary in low resource African countries to ensure effective and equitable use of health information technology and to realize national eHealth goals such as interoperability, adoption of standards and data integration. eHealth regulatory frameworks are under-developed in low resource settings, which hampers the progression towards coherent and effective national health information systems. Ontologies have the potential to clarify issues around interoperability and the effectiveness of different standards to deal with different aspects of interoperability. Ontologies can facilitate drafting, reusing, implementing and compliance testing of eHealth regulations. In this regard, we have developed an OWL ontology to capture key concepts and relations concerning interoperability and standards. The ontology includes an operational definition for interoperability and is an initial step towards the development of a knowledge representation modeling platform for eHealth regulation and governance.},
year = {2014},
journal = {Foundations of Health Information Engineering and Systems, Revised and Selected Papers, Lecture Notes in Computer Science Volume 7789},
pages = {107-124},
month = {15/09},
isbn = {978-3-642-53955-8},
}
Representing temporal data in conceptual data models and ontologies is required by various application domains. For it to be useful for modellers to represent the information precisely and reason over it, it is essential to have a language that is expressive enough to capture the required operational semantics of the time-varying information. Temporal modelling languages have little support for temporal attributes, if at all, yet attributes are a standard element in the widely used conceptual modelling languages such as EER and UML. This hiatus prevents one to utilise a complete temporal conceptual data model and keep track of evolving values of data and its interaction with temporal classes. A rich axiomatisation of fully temporised attributes is possible with a minor extension to the already very expressive description logic language DLRUS. We formalise the notion of transition of attributes, and their interaction with transition of classes. The transition specified for attributes are extension, evolution, and arbitrary quantitative extension.
@misc{91,
author = {Nasubo Ongoma and Maria Keet and Tommie Meyer},
title = {Transition Constraints for Temporal Attributes},
abstract = {Representing temporal data in conceptual data models and ontologies
is required by various application domains. For it to be useful for modellers to
represent the information precisely and reason over it, it is essential to have a language
that is expressive enough to capture the required operational semantics of
the time-varying information. Temporal modelling languages have little support
for temporal attributes, if at all, yet attributes are a standard element in the widely
used conceptual modelling languages such as EER and UML. This hiatus prevents
one to utilise a complete temporal conceptual data model and keep track of
evolving values of data and its interaction with temporal classes. A rich axiomatisation
of fully temporised attributes is possible with a minor extension to the
already very expressive description logic language DLRUS. We formalise the
notion of transition of attributes, and their interaction with transition of classes.
The transition specified for attributes are extension, evolution, and arbitrary quantitative
extension.},
year = {2014},
journal = {27th International Workshop on Description Logics},
month = {17/07 - 20/07},
url = {http://ceur-ws.org/Vol-1193/paper_25.pdf},
}
The preferential approach to nonmonotonic reasoning was consolidated in depth by Krause, Lehmann and Magidor (KLM) for propositional logic in the early 90's. In recent years, there have been efforts to extend their framework to Description Logics (DLs) and a solid (though preliminary) theoretical foundation has already been established towards this aim. Despite this foundation, the generalisation of the propositional framework to DLs is not yet complete and there are multiple proposals for entailment in this context with no formal system for deciding between these. In addition, there are virtually no existing preferential reasoning implementations to speak of for DL-based ontologies. The goals of this PhD are to provide a complete generalisation of the preferential framework of KLM to the DL ALC, provide a formal understanding of the relationships between the multiple proposals for entailment in this context, and finally, to develop an accompanying defeasible reasoning system for DL-based ontologies with performance that is suitable for use in existing ontology development settings.
@{89,
author = {Tommie Meyer and Kody Moodley and U. Sattler},
title = {Practical Defeasible Reasoning for Description Logics},
abstract = {The preferential approach to nonmonotonic reasoning was consolidated in depth by Krause, Lehmann and Magidor (KLM) for propositional logic in the early 90's. In recent years, there have been efforts to extend their framework to Description Logics (DLs) and a solid (though preliminary) theoretical foundation has already been established towards this aim. Despite this foundation, the generalisation of the propositional framework to DLs is not yet complete and there are multiple proposals for entailment in this context with no formal system for deciding between these. In addition, there are virtually no existing preferential reasoning implementations to speak of for DL-based ontologies. The goals of this PhD are to provide a complete generalisation of the preferential framework of KLM to the DL ALC, provide a formal understanding of the relationships between the multiple proposals for entailment in this context, and finally, to develop an accompanying defeasible reasoning system for DL-based ontologies with performance that is suitable for use in existing ontology development settings.},
year = {2014},
journal = {European Starting AI Researcher Symposium},
pages = {191-200},
month = {18/08-19/08},
address = {Prague, Czech Republic},
isbn = {978-1-61499-421-3},
}
