Theoretical study and empirical investigation of sentence analogies
- Afantenos, Stergos, Lim, Suryani, Prade, Henri, Richard, Gilles
- Authors: Afantenos, Stergos , Lim, Suryani , Prade, Henri , Richard, Gilles
- Date: 2022
- Type: Text , Conference paper
- Relation: 1st Workshop on the Interactions between Analogical Reasoning and Machine Learning at 31st International Joint Conference on Artificial Intelligence - 25th European Conference on Artificial Intelligence, IARML@IJCAI-ECAI 2022, Vienna, Austria, 23 July 2022, CEUR Workshop Proceedings Vol. 3174, p. 15-28
- Full Text:
- Reviewed:
- Description: Analogies between 4 sentences, “a is to b as c is to d”, are usually defined between two pairs of sentences (a, b) and (c, d) by constraining a relation R holding between the sentences of the first pair, to hold for the second pair. From a theoretical perspective, three postulates define an analogy - one of which is the “central permutation” postulate which allows the permutation of central elements b and c. This postulate is no longer appropriate in sentence analogies since the existence of R offers no guarantee in general for the existence of some relation S such that S also holds for the pairs (a, c) and (b, d). In this paper, the “central permutation” postulate is replaced by a weaker “internal reversal” postulate to provide an appropriate definition of sentence analogies. To empirically validate the aforementioned postulate, we build a LSTM as well as baseline Random Forest models capable of learning analogies based on quadruplets. We use the Penn Discourse Treebank (PDTB), the Stanford Natural Language Inference (SNLI) and the Microsoft Research Paraphrase (MSRP) corpora. Our experiments show that our models trained on samples of analogies between (a, b) and (c, d), recognize analogies between (b, a) and (d, c) when the underlying relation is symmetrical, validating thus the formal model of sentence analogies using “internal reversal” postulate. © 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
- Authors: Afantenos, Stergos , Lim, Suryani , Prade, Henri , Richard, Gilles
- Date: 2022
- Type: Text , Conference paper
- Relation: 1st Workshop on the Interactions between Analogical Reasoning and Machine Learning at 31st International Joint Conference on Artificial Intelligence - 25th European Conference on Artificial Intelligence, IARML@IJCAI-ECAI 2022, Vienna, Austria, 23 July 2022, CEUR Workshop Proceedings Vol. 3174, p. 15-28
- Full Text:
- Reviewed:
- Description: Analogies between 4 sentences, “a is to b as c is to d”, are usually defined between two pairs of sentences (a, b) and (c, d) by constraining a relation R holding between the sentences of the first pair, to hold for the second pair. From a theoretical perspective, three postulates define an analogy - one of which is the “central permutation” postulate which allows the permutation of central elements b and c. This postulate is no longer appropriate in sentence analogies since the existence of R offers no guarantee in general for the existence of some relation S such that S also holds for the pairs (a, c) and (b, d). In this paper, the “central permutation” postulate is replaced by a weaker “internal reversal” postulate to provide an appropriate definition of sentence analogies. To empirically validate the aforementioned postulate, we build a LSTM as well as baseline Random Forest models capable of learning analogies based on quadruplets. We use the Penn Discourse Treebank (PDTB), the Stanford Natural Language Inference (SNLI) and the Microsoft Research Paraphrase (MSRP) corpora. Our experiments show that our models trained on samples of analogies between (a, b) and (c, d), recognize analogies between (b, a) and (d, c) when the underlying relation is symmetrical, validating thus the formal model of sentence analogies using “internal reversal” postulate. © 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
Significance level of a big data query by exploiting business processes and strategies
- Dinh, Loan, Karmakar, Gour, Kamruzzaman, Joarder, Stranieri, Andrew
- Authors: Dinh, Loan , Karmakar, Gour , Kamruzzaman, Joarder , Stranieri, Andrew
- Date: 2018
- Type: Text , Conference paper
- Relation: 13th Joint International Baltic Conference on Databases and Information Systems Forum and Doctoral Consortium, Baltic-DB and IS Forum-DC 2018 Vol. 2158, p. 63-73
- Full Text: false
- Reviewed:
- Description: Querying data is one of the most frequent activities in business organisations. The tasks involving queries for big data collection, extraction and analysis have never been easy, because to obtain the high quality responses, the expected outcome from these tasks need to be more accurate and highly relevant to a business organisation. The emergence of big data era has further complicated the task. The enormous volume of data from diverse sources and the variety of queries impose a big challenge on business organisations on how to extract deep insight from big data within acceptable time. Determining significance levels of queries based on their relevance to business organisations is able to deal with such challenge. To address this issue, up to our knowledge, there exists only one approach in the literature to calculate the significance level of a query. However, in this approach, only business processes are considered by manually selecting weights for core and non-core business processes. As the significance level of a query must express the importance of that query to a business organisation, it has to be calculated based on the consideration of business strategic direction, which requires the consideration of both business processes and strategies. This paper proposes an approach for the first time where the significance level of a query is determined by exploiting process contributions and strategy priorities. The results produced by our proposed approach using a business case study show the queries that are associated with more important business processes and higher priority strategies have higher significance levels. This vindicates the application of the significance level in a query to dynamically scale the semantic information use in capturing the appropriate level of deep insight and relevant information required for a business organisation. Copyright © 2018 for this paper by the papers' authors.
Engaging communities for prioritising natural resource management and biodiversity conservation actions
- Milne, Robert, Hansen, Birgita
- Authors: Milne, Robert , Hansen, Birgita
- Date: 2016
- Type: Text , Conference proceedings
- Relation: 3rd Annual Conference of Research@Locate, R@Loc 2016; Melbourne, Australia; 12th-14th April 2016; published in CEUR Workshop Proceedings
- Full Text:
- Description: Citizen science has a significant contribution to Natural Resource Management (NRM) through the acquisition and sharing of knowledge. Innovative online technology is playing an increasing role in the support and implementation of citizen science activities. Two projects being conducted in Victoria are using web-based spatial applications to facilitate and support the use of community sourced information for natural resource management and biodiversity conservation. The Natural Resource Management planning portal (NRMPP) is a regional catchment planning tool designed for Catchment Management Authorities and Landcare organisations to plan and prioritise natural resource management works. State-wide Flora and Fauna Teams (SWIFFT) is network of community groups, individuals and organisations that is underpinned by online technology to share knowledge and data on biodiversity issues throughout Victoria. Open source web-based spatial platforms are being used to deliver existing data from multiple sources, provide tools for the entry of spatial data and to provide information required for decision making. The focus of the two projects is to build knowledge management systems with tools that can be used by the community, land managers and other stakeholders to manage, control and share their own data in an online environment. Submission and sharing of community biodiversity and NRM data using online spatial platforms, and federating it with regional, state and national environmental data is facilitating community engagement and providing a process for identifying opportunities to collaborate on NRM activities and biodiversity conservation projects.
- Authors: Milne, Robert , Hansen, Birgita
- Date: 2016
- Type: Text , Conference proceedings
- Relation: 3rd Annual Conference of Research@Locate, R@Loc 2016; Melbourne, Australia; 12th-14th April 2016; published in CEUR Workshop Proceedings
- Full Text:
- Description: Citizen science has a significant contribution to Natural Resource Management (NRM) through the acquisition and sharing of knowledge. Innovative online technology is playing an increasing role in the support and implementation of citizen science activities. Two projects being conducted in Victoria are using web-based spatial applications to facilitate and support the use of community sourced information for natural resource management and biodiversity conservation. The Natural Resource Management planning portal (NRMPP) is a regional catchment planning tool designed for Catchment Management Authorities and Landcare organisations to plan and prioritise natural resource management works. State-wide Flora and Fauna Teams (SWIFFT) is network of community groups, individuals and organisations that is underpinned by online technology to share knowledge and data on biodiversity issues throughout Victoria. Open source web-based spatial platforms are being used to deliver existing data from multiple sources, provide tools for the entry of spatial data and to provide information required for decision making. The focus of the two projects is to build knowledge management systems with tools that can be used by the community, land managers and other stakeholders to manage, control and share their own data in an online environment. Submission and sharing of community biodiversity and NRM data using online spatial platforms, and federating it with regional, state and national environmental data is facilitating community engagement and providing a process for identifying opportunities to collaborate on NRM activities and biodiversity conservation projects.
Historic urban landscapes and visualising Ballarat : Citizen participation for sustainable urban planning and design
- Murphy, Angela, Dahlhaus, Peter, Thompson, Helen
- Authors: Murphy, Angela , Dahlhaus, Peter , Thompson, Helen
- Date: 2016
- Type: Text , Conference proceedings , Conference paper
- Relation: 3rd Annual Conference of Research@Locate, R@Loc 2016; Melbourne, Australia; 12th-14th April 2016; published in CEUR Workshop Proceedings
- Full Text:
- Description: Technological innovation has provided enhanced capacity for knowledge building, for connection and for improved infrastructure planning in the development of the modern city. In parallel to the building of technology supported urban planning and design capacity, a debate has emerged around the need to maximise citizen participation in urban planning. The role of identity, culture and social context has been assessed as being as integral to sustainability in urban planning as is infrastructure management. In 2011 UNESCO, through the mechanism of the recommendation for Historic Urban Landscapes (HUL), created an imperative for the overt recognition of the role of culture, place and identity in sustainable urban planning. The City of Ballarat, Victoria, was the first of a series of international cities to pilot HUL and commit to inclusive citizen based collaboration in urban planning. Through online technology, a platform for partnership building was established. Developed and supported through the Centre for eResearch and Digital Innovation at Federation University Australia, the HUL and Visualising Ballarat portals track the diversity of urban landscapes-from built environment to geomorphology to cultural identity-and facilitate their inclusion in planning and resource allocation. Crowdsourcing was promoted as pivotal in this process, while spatial innovation provided a means through which to bring to life the notion of distinctiveness, identity and place. Through mapping intangibles across complex and diverse groups within community, the potential for improving the quality and management of the planning process was found to be enhanced. Local Area Planning provided a mechanism for a conceptual alignment of past and present and the voice of community has gained a stronger (and more disruptive) voice in determining what communities' value within their lived environment. This shift was assessed as playing an important, and increasingly recognised, role in sustainable urban planning and design.
- Description: CEUR Workshop Proceedings
- Authors: Murphy, Angela , Dahlhaus, Peter , Thompson, Helen
- Date: 2016
- Type: Text , Conference proceedings , Conference paper
- Relation: 3rd Annual Conference of Research@Locate, R@Loc 2016; Melbourne, Australia; 12th-14th April 2016; published in CEUR Workshop Proceedings
- Full Text:
- Description: Technological innovation has provided enhanced capacity for knowledge building, for connection and for improved infrastructure planning in the development of the modern city. In parallel to the building of technology supported urban planning and design capacity, a debate has emerged around the need to maximise citizen participation in urban planning. The role of identity, culture and social context has been assessed as being as integral to sustainability in urban planning as is infrastructure management. In 2011 UNESCO, through the mechanism of the recommendation for Historic Urban Landscapes (HUL), created an imperative for the overt recognition of the role of culture, place and identity in sustainable urban planning. The City of Ballarat, Victoria, was the first of a series of international cities to pilot HUL and commit to inclusive citizen based collaboration in urban planning. Through online technology, a platform for partnership building was established. Developed and supported through the Centre for eResearch and Digital Innovation at Federation University Australia, the HUL and Visualising Ballarat portals track the diversity of urban landscapes-from built environment to geomorphology to cultural identity-and facilitate their inclusion in planning and resource allocation. Crowdsourcing was promoted as pivotal in this process, while spatial innovation provided a means through which to bring to life the notion of distinctiveness, identity and place. Through mapping intangibles across complex and diverse groups within community, the potential for improving the quality and management of the planning process was found to be enhanced. Local Area Planning provided a mechanism for a conceptual alignment of past and present and the voice of community has gained a stronger (and more disruptive) voice in determining what communities' value within their lived environment. This shift was assessed as playing an important, and increasingly recognised, role in sustainable urban planning and design.
- Description: CEUR Workshop Proceedings
Local n-grams for author identification: Notebook for PAN at CLEF 2013 C3 - CEUR Workshop Proceedings
- Layton, Robert, Watters, Paul, Dazeley, Richard
- Authors: Layton, Robert , Watters, Paul , Dazeley, Richard
- Date: 2013
- Type: Text , Conference proceedings
- Full Text:
- Description: Our approach to the author identification task uses existing authorship attribution methods using local n-grams (LNG) and performs a weighted ensemble. This approach came in third for this year's competition, using a relatively simple scheme of weights by training set accuracy. LNG models create profiles, consisting of a list of character n-grams that best represent a particular author's writing. The use of a weighted ensemble improved upon the accuracy of the method without reducing the speed of the algorithm; the submitted solution was not only near the top of the leaderboard in terms of accuracy, but it was also one of the faster algorithms submitted.
- Authors: Layton, Robert , Watters, Paul , Dazeley, Richard
- Date: 2013
- Type: Text , Conference proceedings
- Full Text:
- Description: Our approach to the author identification task uses existing authorship attribution methods using local n-grams (LNG) and performs a weighted ensemble. This approach came in third for this year's competition, using a relatively simple scheme of weights by training set accuracy. LNG models create profiles, consisting of a list of character n-grams that best represent a particular author's writing. The use of a weighted ensemble improved upon the accuracy of the method without reducing the speed of the algorithm; the submitted solution was not only near the top of the leaderboard in terms of accuracy, but it was also one of the faster algorithms submitted.
- «
- ‹
- 1
- ›
- »