Kristin Stock | Massey University (original) (raw)

Papers by Kristin Stock

Trends in ecology & evolution, Feb 1, 2024

Earthquake-prone countries are exploring earthquake early warning (EEW) systems as a risk mitigat... more Earthquake-prone countries are exploring earthquake early warning (EEW) systems as a risk mitigation measure. However, establishing a comprehensive EEW system would require a substantial financial investment, and for many countries, such systems are not economically viable. For Aotearoa New Zealand, with a population of just under five million people, appropriating significant financial investments towards development of an EEW system cost-effectively is likely to be challenging. This research project, launched in February 2020, explores the feasibility of a socio-technical EEW solution for Aotearoa New Zealand. An EEW system may be viable through interconnecting low-cost sensors and recorders through existing communication infrastructures. The project explores the possibility of utilising emerging internet-of-things (IoT) technologies such as micro-electromechanical systems (MEMS) embedded sensors. The sensors may have lower sensitivity and coarse recording systems. Hence, to operationalise this approach it may require a denser network of sensors to achieve an acceptable level of reliability and also rely on the participation and acceptance of engaged citizens. The project seeks to answer the research question: Is it feasible to form an EEW system through a community-engaged network of low-cost sensors? The project is conducting two initial concurrent phases to explore the social and technical challenges and opportunities: • Phase-1: Community-of-practice development and engagement with various communities to scope the challenges and opportunities of establishing an EEW system. • Phase-2: Explore and examine the opportunities, capabilities, challenges and limitations of developing an earthquake early warning system and applications driven by a network of off-the-shelf MEMS devices and IoT infrastructure. This poster shows the project’s progress-to-date on these two phases. The poster also outlines planned activities and expected outputs

The analysis follows the standard two-stage process of structural equational modelling (Hair et a... more The analysis follows the standard two-stage process of structural equational modelling (Hair et al., 2014). First, we conducted a measurement model assessment through factor analysis. Second, we conducted structural model assessment; evaluating the causal relationships of the usability factors to the dependent variable continued intention to use.

GI_Forum

Vast numbers of biological specimens (e.g. flora, fauna, soils) are stored in collections globall... more Vast numbers of biological specimens (e.g. flora, fauna, soils) are stored in collections globally. Many of these have only a natural-language location description, such as '200ft above and south of main highway, 1.1 miles west of Porters Pass', and numerical coordinates are unknown. The BioWhere project is pioneering methods to automatically determine the geographic coordinates (georeferences) of complex location descriptions. Particular challenges are posed by the variable accuracy of recent and historical data that might be used to train models to predict geographic coordinates from the natural-language descriptions; by the presence of historical place names in the descriptions that are not stored in existing gazetteers; and by the vague and context-sensitive nature (e.g. above, on, south of) of the descriptions. We are addressing these challenges by extending the latest transformer-based deep learning models to parse locality descriptions, and to build models for specific spatial terms that incorporate geographic context and data quality to more accurately predict georeferences. We also describe a gazetteer that contains enriched cultural content to support georeferencing of historical records, and to serve as a store of New Zealand Māori cultural knowledge for future generations.

Frontiers in Communication

Earthquake early warning (EEW) systems aim to warn end-users of incoming ground shaking from eart... more Earthquake early warning (EEW) systems aim to warn end-users of incoming ground shaking from earthquakes that have ruptured further afield, potentially reducing risks to lives and properties. EEW is a socio-technical system involving technical and social processes. This paper contributes to advancing EEW research by conducting a literature review investigating the social science knowledge gap in EEW systems. The review of 70 manuscripts found that EEW systems could benefit society, and the benefits may go beyond its direct function for immediate earthquake response. The findings also show that there are social processes involved in designing, developing, and implementing people-centered EEW systems. Therefore, social science research should not just be concerned with the end-user response but also investigate various stakeholders' involvement throughout the development process of EEW systems. Additionally, EEW is a rapidly evolving field of study, and social science research mus...

Zenodo (CERN European Organization for Nuclear Research), Dec 8, 2021

In this paper, we investigate sense of place in the context of favourite places, exploring the re... more In this paper, we investigate sense of place in the context of favourite places, exploring the reasons people give for preferring their favourite places over other places. We conducted an online survey in which we asked 114 respondents to tell us about their favourite places in New Zealand, through textual descriptions and specific, structured questions. Our results show that favourite places are most strongly preferred for their attractiveness, their intrinsic value, and the feelings of safety they engender. Economic value and genealogical links were least important in place preference. Beach environments were also given as common reasons for place preference, and activities were an important factor, with people mentioning friends and family, weather and recreational pursuits such as walking and beach activities. Our analysis also showed correlation between place attachment, identification and spiritual connection for favourite places.

Critical Infrastructures (CI) such as electricity, water, fuel, telecommunication, and road netwo... more Critical Infrastructures (CI) such as electricity, water, fuel, telecommunication, and road networks are a crucial factor for secure and reliable operation of a society. In a non-emergency situation, most businesses operate on an individual infrastructure. However, after major natural disasters such as earthquakes, the conflicts and complex interdependencies among the different infrastructures can cause significant disturbances, as a failure can propagate from one infrastructure to another. To overcome these concerns, this study has developed a Knowledge-Centered Decision Support System (KCDSS) through an integrated impact assessment framework to model CI interdependencies. The KCDSS can predict the CIs’ dynamic behaviour, as well as their reliability and flexibility in response to large-scale disturbances. The KCDSS can be used as an interface between risk assessment and economic modelling tools to quantify the economic consequences resulting from the infrastructure damage. The KCDSS can show the overall level of service (LOS) of a selected region using the information about damage to infrastructure components. The output of the KCDSS are shown through Geographical Information System (GIS) based time stamped outage maps, which can be used by emergency management stakeholders to decisions about the various impacts of infrastructure failure and examine post-disaster recovery options including reinforcement planning, identification of vulnerabilities, and adding or discarding redundancies in an infrastructure network

International Journal of Disaster Risk Reduction, 2020

Mobile apps have the potential to aid disaster response by providing an avenue to distribute rele... more Mobile apps have the potential to aid disaster response by providing an avenue to distribute relevant and timecritical information to the public. Disaster apps already exist in the app markets. However, it is a challenge to engage users in retaining disaster apps on their smartphones. A mixed-methods approach is used in this study to investigate whether usability factors affect users' intention to continue to use an app (referred to here as continuance intention). First, quantitative methods, applying structural equation modelling with survey data from 271 disaster app users, tested a usability-continuance model. Second, a qualitative usability inquiry, using in-depth interviews with 18 participants, explored the users' insights of the relationships identified from the quantitative modelling. The results showed five usability factors to have significant influence on continuance intention. The key positive influencers are (1) users' perceptions as to whether the app delivers its function (app utility); (2) whether it does so dependably (app dependability); and (3) whether it presents information that can be easily understood (user-interface output). Subsequently, too much focus on (4) user-interface graphics and (5) userinterface input can discourage continuance intention. The insights from the qualitative inquiry provide further meaning and context to these relationships. The results have practical implications for designers and developers, guiding what factors to focus on to enhance the continuance intention of disaster apps.

Social Network Analysis and Mining, Dec 11, 2022

Spatial Cognition & Computation

Cluster 1: overlaps (32, 33) Does not justify separate sense as only geometry type differs from S... more Cluster 1: overlaps (32, 33) Does not justify separate sense as only geometry type differs from Sense 1. Cluster 2: across some other object, indicated by nearness diagrams (29, 30) Sense 2: Objects that are across some other object from Cluster 3: covering (43, 48, 51) Sense 3: Objects that are covering (multiple) Through Core of preposition: overlaps (35, 39, 33) Sense 1: Objects that are overlapping Cluster 1: polygon geometries Does not justify separate sense as only geometry type differs from Sense 1. Cluster 2: linear geometries Does not justify separate sense as only geometry type differs from Sense 1. Over Core of preposition: overlaps (39) Sense 1: objects are overlapping/crossing Sense 3: overlap + alignment Cluster 1: mainly dominated by overlap using linear and polygon objects Does not justify separate sense as only geometry type differs from Sense 1 Cluster 2: emphasis on verticality, often polygon/point like objects that sit in a vertically dominant position, so more like one object on top of (or nearly on top of) another Sense 2: One object is above another object Cluster 3: pairs of linear objects (whether aligned or not aligned) Does not justify separate sense as only geometry type differs from Sense 1 Adjacent to Core of preposition: touches (all senses) (36), semantically similar proximity also important Sense 1: objects are touching or nearly touching Cluster 1: overlaps (32) Sense 2: there is some overlap in the objects (with vague boundaries)probably not actually an extra sense, just a stretching of the main sense Cluster 2: linear features, proximity and touching Does not justify separate sense as only geometry type differs from Sense 1 Cluster 3: multiple objects-sides of (50) Does not justify separate sense as only the frequency differs from Sense 1 Beside Core of preposition: touching relation (all three clusters), proximity also important Sense 1: objects are touching or close to each other Cluster 1: closeness and touching Sense 1: objects are touching or close to each other Cluster 2: close and touching, line and polygon Does not justify separate sense as only geometry type differs from Sense 1 Cluster 3: close and parallel, line types Sense 2: objects are close, linear, and parallel Line types alone doesn't justify separate sense, but parallelism does Close to Core of preposition: no three-way core, proximity, touching less important Sense 1: objects are close to each other Cluster 1: polygons, close to each other, but mostly not touching Sense 1: objects are close to each other Cluster 2: linear, parallel most important, but other orientations also permitted Linear parallelism not so strong as for beside (other orientations score more highly), so does not justify separate sense Cluster 3: no separate sense Same as sense 1 Near Core of preposition: no three-way core, proximity, touching even less important than for close to Sense 1: near (proximity) Only one sense Cluster 1: proximity, 3 and 53 are for expressions that involve parts (eastern part, centre part), 28 is disjoint (similar to proximity) Sense 1: near (proximity) Cluster 2 (only 2 expressions): proximity, for line-polygon pairs Does not justify separate sense as only geometry type differs from Sense 1 Cluster 3 (only 2 expressions): touches and proximity for pairs of polygons Does not justify separate sense, for some reason diagrams showing a person are selected Next to Core of preposition: proximity most important, touching close second Sense 1: proximity or touching between objects Cluster 1: touching most important, proximity second, polygon pairs Sense 1: proximity or touching between objects Cluster 2: proximity, 49 sometimes used to indicate linear object (a nature reserve located next to the coast), sometimes to describe part of an object (the side of the garden next to main street), some degree of overlap, linear and parallel Sense 2: linear parallelism Diagram 49 sometimes used to indicate linear object Outside Core of preposition: proximity most important, touching much lower Sense 1: proximity of objects Cluster 1: proximity, pairs of polygons, multiple objects Does not justify separate sense as objects are either multiple or have different geometries Cluster 2: proximity, overlaps (a trail outside of the park), line and polygon Sense 2: objects have partial overlap Off Core of preposition: no diagrams in all four clusters, proximity most dominant Sense 1: proximity of objects Cluster 1: proximity, or touching Sense 1: proximity of objects Cluster 2: two linear objects, all sorts of relations, but more involving touching/overlapping Sense 2: Overlapping of linear objects Cluster 3: linear objects overlapping or touching Does not justify separate sense as only geometry type differs Sense 2 Cluster 4: proximity, includes multiple object types Does not justify separate sense as objects are either multiple or have different geometry types Past Core of preposition: proximity, but no diagram is in all three clusters Sense 1: proximity (includes by the side of) Cluster 1: proximity, or touching (less important) Sense 1: proximity (includes by the side of) Cluster 2: with verb, bounded by, flanked by or, by the side of (probably two senses) Sense 2: enclosure (with appropriate verb) Cluster 3: proximity, multiple objects Does not justify separate sense as objects are multiple (sense 1) By Core of preposition: proximity, but no diagram is in all three clusters Sense 1: proximity (includes by the side of) Cluster 1: proximity, or touching (less important) Sense 1: proximity (includes by the side of) Cluster 2: with verb, bounded by, flanked by or by the side of (probably two senses) Sense 2: enclosure (with appropriate verb) Cluster 3: proximity, multiple objects Does not justify separate sense as objects are multiple (sense 1) Beyond Core of preposition: no overlap Sense 1 (only sense): proximity, with locatum on the other side of relatum from implied observer position Cluster 1: proximity, but implies observer position (diagram 2) Does not justify separate sense as objects are close with a presence of an observer Cluster 2: touching lines, again implies observer position (diagram 9) Does not justify separate sense as linear objects are close with a presence of an observer

Spatial Cognition & Computation

New Zealand Society for Earthquake Engineering, Apr 14, 2021

International Journal of Spatial Data Infrastructures Research, 2010

A number of efforts have been made in recent years to define standards for the description of res... more A number of efforts have been made in recent years to define standards for the description of resources (including web services) in services oriented architectures. These standards often use description logic ontologies (for example, OWLS) and are intended to be machine-readable. They have been applied to geospatial web services to describe the functions that those services perform in a way that can be automatically interpreted by systems. By contrast, little effort has gone into the development of human readable descriptions of resources in a services oriented architecture, other than using unstructured natural language. e-Framework is an infrastructure for the higher education environment that provides a typology of human-readable artefacts that can be used to describe resources, and provides an internal structure for those artefacts. e-Framework has thus far not been used with geospatial information even though geospatial information has a number of important roles in education and research, and has a well-organised community of users and creators. This paper applies the e-Framework infrastructure to OGC web services, and also recommends the refinement of e-Framework with the use of the SUMO Upper Level Ontology to define Service Genres, the most abstract level of artefacts in e-Framework. It then illustrates the ways in which the Open Geospatial Consortium standards and specifications may be described in e-Framework. The work evaluates SUMO for e-Framework purposes, finding that its use for Service Genres is possible and offers a number of gains. It also

Spatial relations are one of the most important components in a location description, conveying i... more Spatial relations are one of the most important components in a location description, conveying information about proximity, direction, adjacency and topology among other things. However, despite being studied for many years, the semantics of spatial relations are still not well understood,particularly given that the use of spatial relations can vary with context. In this paper we investigate whether it is possible to mine the semantics of spatial relations from text, particularly focusing on semantic similarity, but also exploring the extraction of richer semantic informationabout the relationships between spatial relations, with the long term goal of moving towards the automation of the interpretation and generation of locative expressions. We test three similarity methods, including a bag of words technique, with both general and geospatial corpora, andusing word embeddings. We compare the results to ground truth data from human subjects experiments.

International Journal of Geographical Information Science, 2021

Dataset and code used in a journal paper entitled Detecting Geospatial Location Descriptions in N... more Dataset and code used in a journal paper entitled Detecting Geospatial Location Descriptions in Natural Language Text, published in the International Journal of Geographical Information Science. Abstract:<br>References to geographic locations are common in text data sources including social media and web pages. They take different forms, from simple place names to relative expressions that describe location through a spatial relationship to a reference object (e.g. the house beside the Waikato River). Often complex, multi-word phrases are employed (e.g. the road and railway cross at right angles; the road in line with the canal) where spatial relationships are communicated with various parts of speech including prepositions, verbs, adverbs and adjectives. We address the problem of automatically detecting relative geospatial location descriptions, which we define as those that include spatial relation terms referencing geographic objects, and distinguishing them from non-geographical descriptions of location (e.g. the book on the table). We experiment with several methods for automated classification of text expressions, using features for machine learning that include bag of words that detect distinctive words; word embeddings that encode meanings of words; and manually identified language patterns that characterise geospatial expressions. Using three data sets created for this study, we find that ensemble and meta-classifier approaches, that variously combine predictions from several other classifiers with data features, provide the best F-measure of 0.90 for detecting geospatial expressions. <br>

Trends in ecology & evolution, Feb 1, 2024

Earthquake-prone countries are exploring earthquake early warning (EEW) systems as a risk mitigat... more Earthquake-prone countries are exploring earthquake early warning (EEW) systems as a risk mitigation measure. However, establishing a comprehensive EEW system would require a substantial financial investment, and for many countries, such systems are not economically viable. For Aotearoa New Zealand, with a population of just under five million people, appropriating significant financial investments towards development of an EEW system cost-effectively is likely to be challenging. This research project, launched in February 2020, explores the feasibility of a socio-technical EEW solution for Aotearoa New Zealand. An EEW system may be viable through interconnecting low-cost sensors and recorders through existing communication infrastructures. The project explores the possibility of utilising emerging internet-of-things (IoT) technologies such as micro-electromechanical systems (MEMS) embedded sensors. The sensors may have lower sensitivity and coarse recording systems. Hence, to operationalise this approach it may require a denser network of sensors to achieve an acceptable level of reliability and also rely on the participation and acceptance of engaged citizens. The project seeks to answer the research question: Is it feasible to form an EEW system through a community-engaged network of low-cost sensors? The project is conducting two initial concurrent phases to explore the social and technical challenges and opportunities: • Phase-1: Community-of-practice development and engagement with various communities to scope the challenges and opportunities of establishing an EEW system. • Phase-2: Explore and examine the opportunities, capabilities, challenges and limitations of developing an earthquake early warning system and applications driven by a network of off-the-shelf MEMS devices and IoT infrastructure. This poster shows the project’s progress-to-date on these two phases. The poster also outlines planned activities and expected outputs

The analysis follows the standard two-stage process of structural equational modelling (Hair et a... more The analysis follows the standard two-stage process of structural equational modelling (Hair et al., 2014). First, we conducted a measurement model assessment through factor analysis. Second, we conducted structural model assessment; evaluating the causal relationships of the usability factors to the dependent variable continued intention to use.

GI_Forum

Vast numbers of biological specimens (e.g. flora, fauna, soils) are stored in collections globall... more Vast numbers of biological specimens (e.g. flora, fauna, soils) are stored in collections globally. Many of these have only a natural-language location description, such as '200ft above and south of main highway, 1.1 miles west of Porters Pass', and numerical coordinates are unknown. The BioWhere project is pioneering methods to automatically determine the geographic coordinates (georeferences) of complex location descriptions. Particular challenges are posed by the variable accuracy of recent and historical data that might be used to train models to predict geographic coordinates from the natural-language descriptions; by the presence of historical place names in the descriptions that are not stored in existing gazetteers; and by the vague and context-sensitive nature (e.g. above, on, south of) of the descriptions. We are addressing these challenges by extending the latest transformer-based deep learning models to parse locality descriptions, and to build models for specific spatial terms that incorporate geographic context and data quality to more accurately predict georeferences. We also describe a gazetteer that contains enriched cultural content to support georeferencing of historical records, and to serve as a store of New Zealand Māori cultural knowledge for future generations.

Frontiers in Communication

Earthquake early warning (EEW) systems aim to warn end-users of incoming ground shaking from eart... more Earthquake early warning (EEW) systems aim to warn end-users of incoming ground shaking from earthquakes that have ruptured further afield, potentially reducing risks to lives and properties. EEW is a socio-technical system involving technical and social processes. This paper contributes to advancing EEW research by conducting a literature review investigating the social science knowledge gap in EEW systems. The review of 70 manuscripts found that EEW systems could benefit society, and the benefits may go beyond its direct function for immediate earthquake response. The findings also show that there are social processes involved in designing, developing, and implementing people-centered EEW systems. Therefore, social science research should not just be concerned with the end-user response but also investigate various stakeholders' involvement throughout the development process of EEW systems. Additionally, EEW is a rapidly evolving field of study, and social science research mus...

Zenodo (CERN European Organization for Nuclear Research), Dec 8, 2021

In this paper, we investigate sense of place in the context of favourite places, exploring the re... more In this paper, we investigate sense of place in the context of favourite places, exploring the reasons people give for preferring their favourite places over other places. We conducted an online survey in which we asked 114 respondents to tell us about their favourite places in New Zealand, through textual descriptions and specific, structured questions. Our results show that favourite places are most strongly preferred for their attractiveness, their intrinsic value, and the feelings of safety they engender. Economic value and genealogical links were least important in place preference. Beach environments were also given as common reasons for place preference, and activities were an important factor, with people mentioning friends and family, weather and recreational pursuits such as walking and beach activities. Our analysis also showed correlation between place attachment, identification and spiritual connection for favourite places.

Critical Infrastructures (CI) such as electricity, water, fuel, telecommunication, and road netwo... more Critical Infrastructures (CI) such as electricity, water, fuel, telecommunication, and road networks are a crucial factor for secure and reliable operation of a society. In a non-emergency situation, most businesses operate on an individual infrastructure. However, after major natural disasters such as earthquakes, the conflicts and complex interdependencies among the different infrastructures can cause significant disturbances, as a failure can propagate from one infrastructure to another. To overcome these concerns, this study has developed a Knowledge-Centered Decision Support System (KCDSS) through an integrated impact assessment framework to model CI interdependencies. The KCDSS can predict the CIs’ dynamic behaviour, as well as their reliability and flexibility in response to large-scale disturbances. The KCDSS can be used as an interface between risk assessment and economic modelling tools to quantify the economic consequences resulting from the infrastructure damage. The KCDSS can show the overall level of service (LOS) of a selected region using the information about damage to infrastructure components. The output of the KCDSS are shown through Geographical Information System (GIS) based time stamped outage maps, which can be used by emergency management stakeholders to decisions about the various impacts of infrastructure failure and examine post-disaster recovery options including reinforcement planning, identification of vulnerabilities, and adding or discarding redundancies in an infrastructure network

International Journal of Disaster Risk Reduction, 2020

Mobile apps have the potential to aid disaster response by providing an avenue to distribute rele... more Mobile apps have the potential to aid disaster response by providing an avenue to distribute relevant and timecritical information to the public. Disaster apps already exist in the app markets. However, it is a challenge to engage users in retaining disaster apps on their smartphones. A mixed-methods approach is used in this study to investigate whether usability factors affect users' intention to continue to use an app (referred to here as continuance intention). First, quantitative methods, applying structural equation modelling with survey data from 271 disaster app users, tested a usability-continuance model. Second, a qualitative usability inquiry, using in-depth interviews with 18 participants, explored the users' insights of the relationships identified from the quantitative modelling. The results showed five usability factors to have significant influence on continuance intention. The key positive influencers are (1) users' perceptions as to whether the app delivers its function (app utility); (2) whether it does so dependably (app dependability); and (3) whether it presents information that can be easily understood (user-interface output). Subsequently, too much focus on (4) user-interface graphics and (5) userinterface input can discourage continuance intention. The insights from the qualitative inquiry provide further meaning and context to these relationships. The results have practical implications for designers and developers, guiding what factors to focus on to enhance the continuance intention of disaster apps.

Social Network Analysis and Mining, Dec 11, 2022

Spatial Cognition & Computation

Cluster 1: overlaps (32, 33) Does not justify separate sense as only geometry type differs from S... more Cluster 1: overlaps (32, 33) Does not justify separate sense as only geometry type differs from Sense 1. Cluster 2: across some other object, indicated by nearness diagrams (29, 30) Sense 2: Objects that are across some other object from Cluster 3: covering (43, 48, 51) Sense 3: Objects that are covering (multiple) Through Core of preposition: overlaps (35, 39, 33) Sense 1: Objects that are overlapping Cluster 1: polygon geometries Does not justify separate sense as only geometry type differs from Sense 1. Cluster 2: linear geometries Does not justify separate sense as only geometry type differs from Sense 1. Over Core of preposition: overlaps (39) Sense 1: objects are overlapping/crossing Sense 3: overlap + alignment Cluster 1: mainly dominated by overlap using linear and polygon objects Does not justify separate sense as only geometry type differs from Sense 1 Cluster 2: emphasis on verticality, often polygon/point like objects that sit in a vertically dominant position, so more like one object on top of (or nearly on top of) another Sense 2: One object is above another object Cluster 3: pairs of linear objects (whether aligned or not aligned) Does not justify separate sense as only geometry type differs from Sense 1 Adjacent to Core of preposition: touches (all senses) (36), semantically similar proximity also important Sense 1: objects are touching or nearly touching Cluster 1: overlaps (32) Sense 2: there is some overlap in the objects (with vague boundaries)probably not actually an extra sense, just a stretching of the main sense Cluster 2: linear features, proximity and touching Does not justify separate sense as only geometry type differs from Sense 1 Cluster 3: multiple objects-sides of (50) Does not justify separate sense as only the frequency differs from Sense 1 Beside Core of preposition: touching relation (all three clusters), proximity also important Sense 1: objects are touching or close to each other Cluster 1: closeness and touching Sense 1: objects are touching or close to each other Cluster 2: close and touching, line and polygon Does not justify separate sense as only geometry type differs from Sense 1 Cluster 3: close and parallel, line types Sense 2: objects are close, linear, and parallel Line types alone doesn't justify separate sense, but parallelism does Close to Core of preposition: no three-way core, proximity, touching less important Sense 1: objects are close to each other Cluster 1: polygons, close to each other, but mostly not touching Sense 1: objects are close to each other Cluster 2: linear, parallel most important, but other orientations also permitted Linear parallelism not so strong as for beside (other orientations score more highly), so does not justify separate sense Cluster 3: no separate sense Same as sense 1 Near Core of preposition: no three-way core, proximity, touching even less important than for close to Sense 1: near (proximity) Only one sense Cluster 1: proximity, 3 and 53 are for expressions that involve parts (eastern part, centre part), 28 is disjoint (similar to proximity) Sense 1: near (proximity) Cluster 2 (only 2 expressions): proximity, for line-polygon pairs Does not justify separate sense as only geometry type differs from Sense 1 Cluster 3 (only 2 expressions): touches and proximity for pairs of polygons Does not justify separate sense, for some reason diagrams showing a person are selected Next to Core of preposition: proximity most important, touching close second Sense 1: proximity or touching between objects Cluster 1: touching most important, proximity second, polygon pairs Sense 1: proximity or touching between objects Cluster 2: proximity, 49 sometimes used to indicate linear object (a nature reserve located next to the coast), sometimes to describe part of an object (the side of the garden next to main street), some degree of overlap, linear and parallel Sense 2: linear parallelism Diagram 49 sometimes used to indicate linear object Outside Core of preposition: proximity most important, touching much lower Sense 1: proximity of objects Cluster 1: proximity, pairs of polygons, multiple objects Does not justify separate sense as objects are either multiple or have different geometries Cluster 2: proximity, overlaps (a trail outside of the park), line and polygon Sense 2: objects have partial overlap Off Core of preposition: no diagrams in all four clusters, proximity most dominant Sense 1: proximity of objects Cluster 1: proximity, or touching Sense 1: proximity of objects Cluster 2: two linear objects, all sorts of relations, but more involving touching/overlapping Sense 2: Overlapping of linear objects Cluster 3: linear objects overlapping or touching Does not justify separate sense as only geometry type differs Sense 2 Cluster 4: proximity, includes multiple object types Does not justify separate sense as objects are either multiple or have different geometry types Past Core of preposition: proximity, but no diagram is in all three clusters Sense 1: proximity (includes by the side of) Cluster 1: proximity, or touching (less important) Sense 1: proximity (includes by the side of) Cluster 2: with verb, bounded by, flanked by or, by the side of (probably two senses) Sense 2: enclosure (with appropriate verb) Cluster 3: proximity, multiple objects Does not justify separate sense as objects are multiple (sense 1) By Core of preposition: proximity, but no diagram is in all three clusters Sense 1: proximity (includes by the side of) Cluster 1: proximity, or touching (less important) Sense 1: proximity (includes by the side of) Cluster 2: with verb, bounded by, flanked by or by the side of (probably two senses) Sense 2: enclosure (with appropriate verb) Cluster 3: proximity, multiple objects Does not justify separate sense as objects are multiple (sense 1) Beyond Core of preposition: no overlap Sense 1 (only sense): proximity, with locatum on the other side of relatum from implied observer position Cluster 1: proximity, but implies observer position (diagram 2) Does not justify separate sense as objects are close with a presence of an observer Cluster 2: touching lines, again implies observer position (diagram 9) Does not justify separate sense as linear objects are close with a presence of an observer

Spatial Cognition & Computation

New Zealand Society for Earthquake Engineering, Apr 14, 2021

International Journal of Spatial Data Infrastructures Research, 2010

A number of efforts have been made in recent years to define standards for the description of res... more A number of efforts have been made in recent years to define standards for the description of resources (including web services) in services oriented architectures. These standards often use description logic ontologies (for example, OWLS) and are intended to be machine-readable. They have been applied to geospatial web services to describe the functions that those services perform in a way that can be automatically interpreted by systems. By contrast, little effort has gone into the development of human readable descriptions of resources in a services oriented architecture, other than using unstructured natural language. e-Framework is an infrastructure for the higher education environment that provides a typology of human-readable artefacts that can be used to describe resources, and provides an internal structure for those artefacts. e-Framework has thus far not been used with geospatial information even though geospatial information has a number of important roles in education and research, and has a well-organised community of users and creators. This paper applies the e-Framework infrastructure to OGC web services, and also recommends the refinement of e-Framework with the use of the SUMO Upper Level Ontology to define Service Genres, the most abstract level of artefacts in e-Framework. It then illustrates the ways in which the Open Geospatial Consortium standards and specifications may be described in e-Framework. The work evaluates SUMO for e-Framework purposes, finding that its use for Service Genres is possible and offers a number of gains. It also

Spatial relations are one of the most important components in a location description, conveying i... more Spatial relations are one of the most important components in a location description, conveying information about proximity, direction, adjacency and topology among other things. However, despite being studied for many years, the semantics of spatial relations are still not well understood,particularly given that the use of spatial relations can vary with context. In this paper we investigate whether it is possible to mine the semantics of spatial relations from text, particularly focusing on semantic similarity, but also exploring the extraction of richer semantic informationabout the relationships between spatial relations, with the long term goal of moving towards the automation of the interpretation and generation of locative expressions. We test three similarity methods, including a bag of words technique, with both general and geospatial corpora, andusing word embeddings. We compare the results to ground truth data from human subjects experiments.

International Journal of Geographical Information Science, 2021

Dataset and code used in a journal paper entitled Detecting Geospatial Location Descriptions in N... more Dataset and code used in a journal paper entitled Detecting Geospatial Location Descriptions in Natural Language Text, published in the International Journal of Geographical Information Science. Abstract:<br>References to geographic locations are common in text data sources including social media and web pages. They take different forms, from simple place names to relative expressions that describe location through a spatial relationship to a reference object (e.g. the house beside the Waikato River). Often complex, multi-word phrases are employed (e.g. the road and railway cross at right angles; the road in line with the canal) where spatial relationships are communicated with various parts of speech including prepositions, verbs, adverbs and adjectives. We address the problem of automatically detecting relative geospatial location descriptions, which we define as those that include spatial relation terms referencing geographic objects, and distinguishing them from non-geographical descriptions of location (e.g. the book on the table). We experiment with several methods for automated classification of text expressions, using features for machine learning that include bag of words that detect distinctive words; word embeddings that encode meanings of words; and manually identified language patterns that characterise geospatial expressions. Using three data sets created for this study, we find that ensemble and meta-classifier approaches, that variously combine predictions from several other classifiers with data features, provide the best F-measure of 0.90 for detecting geospatial expressions. <br>