2016 49th Hawaii International Conference on System Sciences (HICSS), 2016
An important area of work in big data software engineering involves the design and development of... more An important area of work in big data software engineering involves the design and development of software frameworks for data-intensive systems that perform large-scale data collection and analysis. We report on our work to design and develop a software framework for analyzing the collaborative editing behavior of OpenStreetMap users when working on the task of crisis mapping. Crisis mapping occurs after a disaster or humanitarian crisis and involves the coordination of a distributed set of users who collaboratively work to improve the quality of the map for the impacted area in support of emergency response efforts. Our paper presents the challenges related to the analysis of OpenStreetMap and how our software framework tackles those challenges to enable the efficient processing of gigabytes of OpenStreetMap data. Our framework has already been deployed to analyze crisis mapping efforts in 2015 and has an active development community.
The world wide web has a wealth of information that is related to almost any text classification ... more The world wide web has a wealth of information that is related to almost any text classification task. This paper presents a method for mining the web to improve text classification, by creating a background text set. Our algorithm uses the information gain criterion to create lists of important words for each class of a text categorization problem. It then searches the web on various combinations of these words to produce a set of related data. We use this set of background text with Latent Semantic Indexing classification to create an expanded term by document matrix on which singular value decomposition is done. We provide empirical results that this approach improves accuracy on unseen test examples in different domains.
The world wide web has a wealth of information that is related to almost any text classification ... more The world wide web has a wealth of information that is related to almost any text classification task. This paper presents a method for mining the web to improve text classification, by creating a background text set. Our algorithm uses the information gain criterion to create lists of important words for each class of a text categorization problem. It then searches the web on various combinations of these words to produce a set of related data. We use this set of background text with Latent Semantic Indexing classification to create an expanded term by document matrix on which singular value decomposition is done. We provide empirical results that this approach improves accuracy on unseen test examples in different domains.
Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems - CHI '16, 2016
OpenStreetMap (OSM) is the most widely used volunteer geographic information system. Although it ... more OpenStreetMap (OSM) is the most widely used volunteer geographic information system. Although it is increasingly relied upon during humanitarian response as the most up-todate, accurate, or accessible map of affected areas, the behavior of the mappers who contribute to it is not well understood. In this paper, we explore the work practices and interactions of volunteer mappers operating in the hightempo, high-volume context of disasters. To do this, we built upon and expanded prior network analysis techniques to select high-value portions of the vast OSM data for further qualitative analysis. We then performed detailed content analysis of the identified activity and, where possible, conducted interviews with the participants. This research allowed the identification of seven distinct mapping practices that can be classified according to dimensions of time, space, and interpersonal interaction. Our work represents a baseline for future research about how OSM crisis mapping practices have evolved over time.
2016 49th Hawaii International Conference on System Sciences (HICSS), 2016
An important area of work in big data software engineering involves the design and development of... more An important area of work in big data software engineering involves the design and development of software frameworks for data-intensive systems that perform large-scale data collection and analysis. We report on our work to design and develop a software framework for analyzing the collaborative editing behavior of OpenStreetMap users when working on the task of crisis mapping. Crisis mapping occurs after a disaster or humanitarian crisis and involves the coordination of a distributed set of users who collaboratively work to improve the quality of the map for the impacted area in support of emergency response efforts. Our paper presents the challenges related to the analysis of OpenStreetMap and how our software framework tackles those challenges to enable the efficient processing of gigabytes of OpenStreetMap data. Our framework has already been deployed to analyze crisis mapping efforts in 2015 and has an active development community.
The world wide web has a wealth of information that is related to almost any text classification ... more The world wide web has a wealth of information that is related to almost any text classification task. This paper presents a method for mining the web to improve text classification, by creating a background text set. Our algorithm uses the information gain criterion to create lists of important words for each class of a text categorization problem. It then searches the web on various combinations of these words to produce a set of related data. We use this set of background text with Latent Semantic Indexing classification to create an expanded term by document matrix on which singular value decomposition is done. We provide empirical results that this approach improves accuracy on unseen test examples in different domains.
The world wide web has a wealth of information that is related to almost any text classification ... more The world wide web has a wealth of information that is related to almost any text classification task. This paper presents a method for mining the web to improve text classification, by creating a background text set. Our algorithm uses the information gain criterion to create lists of important words for each class of a text categorization problem. It then searches the web on various combinations of these words to produce a set of related data. We use this set of background text with Latent Semantic Indexing classification to create an expanded term by document matrix on which singular value decomposition is done. We provide empirical results that this approach improves accuracy on unseen test examples in different domains.
Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems - CHI '16, 2016
OpenStreetMap (OSM) is the most widely used volunteer geographic information system. Although it ... more OpenStreetMap (OSM) is the most widely used volunteer geographic information system. Although it is increasingly relied upon during humanitarian response as the most up-todate, accurate, or accessible map of affected areas, the behavior of the mappers who contribute to it is not well understood. In this paper, we explore the work practices and interactions of volunteer mappers operating in the hightempo, high-volume context of disasters. To do this, we built upon and expanded prior network analysis techniques to select high-value portions of the vast OSM data for further qualitative analysis. We then performed detailed content analysis of the identified activity and, where possible, conducted interviews with the participants. This research allowed the identification of seven distinct mapping practices that can be classified according to dimensions of time, space, and interpersonal interaction. Our work represents a baseline for future research about how OSM crisis mapping practices have evolved over time.
Uploads
Papers by Marina Kogan