Benjamin Adams | University of California, Santa Barbara (original) (raw)

Papers by Benjamin Adams

Abstract: Shared reference is an essential aspect of meaning. It is also indispensable for the se... more Abstract: Shared reference is an essential aspect of meaning. It is also indispensable for the semantic web, since it enables to weave the global graph, ie, it allows different users to contribute to an identical referent. For example, an essential kind of referent is a geographic place, to which users may contribute observations. We argue for a human-centric, operational approach towards reference, based on respective human competences.

ABSTRACT Permission to make digital or hard copies of all or part of this work for personal or cl... more ABSTRACT Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

Abstract. Computing user similarity is key for personalized locationbased recommender systems and... more Abstract. Computing user similarity is key for personalized locationbased recommender systems and geographic information retrieval. So far, most existing work has focused on structured or semi-structured data to establish such measures. In this work, we propose topic modeling to exploit sparse, unstructured data, eg, tips and reviews, as an additional feature to compute user similarity.

We have performed a topic classification procedure on a text corpus consisting of the proceedings... more We have performed a topic classification procedure on a text corpus consisting of the proceedings for all ten meetings of the Conference on Spatial Information Theory (COSIT) held between 1992 and 2009, providing a measure with which to answer several kinds of questions about the dynamic conceptual content of that conference series. We have identified topics trending upward and downward, looking particularly at the level of research interest in cultural factors. We have also investigated whether there has been growing interdisciplinarity in the research reported at COSIT meetings by this diverse and dynamic scholarly community. Preliminary results are presented, planned future work is discussed, and additional questions are invited.

Places are more than just a location and spatial footprint. A sense of place is the result of sub... more Places are more than just a location and spatial footprint. A sense of place is the result of subjective experience that a person has from being in a place or from interacting with information about a place. Although it is difficult to directly model a person's conceptualization of sense of place in a computational representation, there exist many natural language data online that describe people's experiences with places and which can be used to learn computational representations. In this paper we evaluate the usage of topic modeling on a set of travel blog entries to identify the themes that are most closely associated with places around the world. Using these representations we can calculate the similarity of places. In addition, by focusing on individual or sets of topics we identify new regions where topics are most salient. Finally we discuss how temporal changes in sense of place can be evaluated using these methods.

When experiencing or describing a new place people will often compare it against other places tha... more When experiencing or describing a new place people will often compare it against other places that they already know. However, this human attention to the simultaneous similarities and differences between places is not reflected in the design of user interfaces of current place search technologies. In this demo, we present Frankenplace, an application for doing similarity-based place search that allows users to interactively find new places based on mixtures of features drawn from different places. The features of places are derived from a combination of authoritative data sources and unstructured observation data from social media, and organized into an extensible set of layers. We demonstrate the Frankenplace interface, which lets a user build a profile of a target place by selecting the most relevant of the properties shared by known places.

Geographic location is a key component for information retrieval on the Web, recommendation syste... more Geographic location is a key component for information retrieval on the Web, recommendation systems in mobile computing and social networks, and placebased integration on the Linked Data cloud. Previous work has addressed how to estimate locations by named entity recognition, from images, and via structured data. In this paper, we estimate geographic regions from unstructured, non geo-referenced text by computing a probability distribution over the Earth's surface. Our methodology combines natural language processing, geostatistics, and a data-driven bottom-up semantics. We illustrate its potential for mapping geographic regions from non geo-referenced text.

ACM GIS, Jan 1, 2011

The semantic integration of heterogeneous, spatiotemporal information is a major challenge for ac... more The semantic integration of heterogeneous, spatiotemporal information is a major challenge for achieving the vision of a multithematic and multi-perspective Digital Earth. The Semantic Web technology stack has been proposed to address the integration problem by knowledge representation languages and reasoning. However approaches such as the Web Ontology Languages (OWL) were developed with decidability in mind. They do not integrate well with established modeling paradigms in the geosciences that are dominated by numerical and geometric methods. Additionally, work on the Semantic Web is mostly feature-centric and a fieldbased view is difficult to integrate. A layer specifying the transition from observation data to classes and relations is missing. In this work we combine OWL with geometric and topological language constructs based on similarity spaces. Our approach provides three main benefits. First, class constructors can be built from a larger palette of mathematical operations based on vector algebra. Second, it affords the representation of prototype-based classes. Third, it facilitates the representation of classes derived from machine learning classifiers that utilize a multi-dimensional feature space. Instead of following a one-size-fits-all approach, our work allows one to derive contextualized OWL ontologies by reification of observation data.

Semantic Web, Jan 1, 2010

One of the key deficiencies of the Semantic Web is its lack of cognitive plausibility. We argue t... more One of the key deficiencies of the Semantic Web is its lack of cognitive plausibility. We argue that by accounting for people's reasoning mechanisms and cognitive representations, the usefulness of information coming from the Semantic Web will be enhanced. More specifically, the utilization and integration of conceptual spaces is proposed as a knowledge representation that affords two important human cognitive mechanisms, i.e., semantic similarity and concept combination. Formal conceptual space algebra serves as the basis for the Conceptual Space Markup Language (CSML), which facilitates the engineering of ontologies using a geometric framework. We demonstrate the usefulness of the approach through a concrete example and suggest directions for future work, especially the need for combining geometric representations and reasoning mechanisms with existing Semantic Web structures.

Geographic Information Science, Jan 1, 2010

Semantic similarity measurement is a key methodology in various domains ranging from cognitive sc... more Semantic similarity measurement is a key methodology in various domains ranging from cognitive science to geographic information retrieval on the Web. Meaningful notions of similarity, however, cannot be determined without taking additional contextual information into account. One way to make similarity measures context-aware is by introducing weights for specific characteristics. Existing approaches to automatically determine such weights are rather limited or require application specific adjustments. In the past, the possibility to tweak similarity theories until they fit a specific use case has been one of the major criticisms for their evaluation. In this work, we propose a novel approach to semi-automatically adapt similarity theories to the user's needs and hence make them context-aware. Our methodology is inspired by the process of georeferencing images in which known control points between the image and geographic space are used to compute a suitable transformation. We propose to semi-automatically calibrate weights to compute inter-instance and inter-concept similarities by allowing the user to adjust pre-computed similarity rankings. These known control similarities are then used to reference other similarity values.

Spatial Information Theory, Jan 1, 2009

The modeling of concepts from a cognitive perspective is important for designing spatial informat... more The modeling of concepts from a cognitive perspective is important for designing spatial information systems that interoperate with human users. Concept representations that are built using geometric and topological conceptual space structures are well suited for semantic similarity and concept combination operations. In addition, concepts that are more closely grounded in the physical world, such as many spatial concepts, have a natural fit with the geometric structure of conceptual spaces. Despite these apparent advantages, conceptual spaces are underutilized because existing formalizations of conceptual space theory have focused on individual aspects of the theory rather than the creation of a comprehensive algebra. In this paper we present a metric conceptual space algebra that is designed to facilitate the creation of conceptual space knowledge bases and inferencing systems. Conceptual regions are represented as convex polytopes and context is built in as a fundamental element. We demonstrate the applicability of the algebra to spatial information systems with a proof-of-concept application.

2009 IEEE International Conference on …, Jan 1, 2009

CSML is a semantic markup language created for the publishing and sharing of conceptual spaces, w... more CSML is a semantic markup language created for the publishing and sharing of conceptual spaces, which are geometric structures that represent semantics at the conceptual level. CSML can be used to describe semantics that are not captured well by the ontology languages commonly used in the Semantic Web. Measurement of the semantic similarity of concepts as well as the combination of concepts without shared properties are common human cognitive tasks. However, these operations present sources of difficulty for tools reliant upon set-theoretic and syllogistic reasoning on symbolic ontologies. In contrast, these operations can be modeled naturally using conceptual spaces. This paper describes the design decisions behind CSML, introduces the key component elements of a CSML document, and presents examples of its usage.