Resolving ambiguous descriptions through visual information

The project “Reference in Discourse” deals with the selection of a specific object from a visual scene in a natural language situation. The goal of this research is to explain this everyday discourse reference task in terms of a concept generation process based on subconceptual visual and verbal information. The system OINC (Object Identification in Natural Communicators) aims at solving this problem in a psychologically adequate way. The system’s difficulties occurring with incomplete and deviant descriptions correspond to the data from experiments with human subjects. The results of these experiments are reported.

The temporal binding deficit hypothesis of autism

Frith has argued that people with autism show “weak central coherence,” an unusual bias toward piecemeal rather than configurational processing and a reduction in the normal tendency to process information in context. However, the precise cognitive and neurological mechanisms underlying weak central coherence are still unknown. We propose the hypothesis that the features of autism associated with weak central coherence result from a reduction in the integration of specialized local neural networks in the brain caused by a deficit in temporal binding. The visuoperceptual anomalies associated with weak central coherence may be attributed to a reduction in synchronization of high-frequency gamma activity between local networks processing local features. The failure to utilize context in language processing in autism can be explained in similar terms. Temporal binding deficits could also contribute to executive dysfunction in autism and to some of the deficits in socialization and communication.

Reading skills are related to global, but not local, acoustic pattern perception

Although reading ability has been related to the processing of simple pitch features such as isolated transitions or continuous modulation spoken language also contains complex patterns of pitch changes that are important for establishing stress location and for segmenting the speech stream. These aspects of spoken language processing depend critically on pitch pattern (global structure) rather than on absolute pitch values (local structure). Here we show that the detection of global structure, and not local structure, is predictive of performance on measures of phonological skill and reading ability, which supports a critical importance of pitch contour processing in the acquisition of literacy.

In search of a theory of list memory: an inquiry into the efects of long-term information on short-term recall

Over recent years, evidence has been accumulating in favour of the importance of long-term information as a variable which can affect the success of short-term recall. Lexicality, word frequency, imagery and meaning have all been shown to augment short term recall performance. Two competing theories as to the causes of this long-term memory influence are outlined and tested in this thesis. The first approach is the order-encoding account, which ascribes the effect to the usage of resources at encoding, hypothesising that word lists which require less effort to process will benefit from increased levels of order encoding, in turn enhancing recall success. The alternative view, trace redintegration theory, suggests that order is automatically encoded phonologically, and that long-term information can only influence the interpretation of the resultant memory trace. The free recall experiments reported here attempted to determine the importance of order encoding as a facilitatory framework and to determine the locus of the effects of long-term information in free recall. Experiments 1 and 2 examined the effects of word frequency and semantic categorisation over a filled delay, and experiments 3 and 4 did the same for immediate recall. Free recall was improved by both long-term factors tested. Order information was not used over a short filled delay, but was evident in immediate recall. Furthermore, it was found that both long-term factors increased the amount of order information retained. Experiment 5 induced an order encoding effect over a filled delay, leaving a picture of short-term processes which are closely associated with long-term processes, and which fit conceptions of short-term memory being part of language processes rather better than either the encoding or the retrieval-based models. Experiments 6 and 7 aimed to determine to what extent phonological processes were responsible for the pattern of results observed. Articulatory suppression affected the encoding of order information where speech rate had no direct influence, suggesting that it is ease of lexical access which is the most important factor in the influence of long-term memory on immediate recall tasks. The evidence presented in this thesis does not offer complete support for either the retrieval-based account or the order encoding account of long-term influence. Instead, the evidence sits best with models that are based upon language-processing. The path urged for future research is to find ways in which this diffuse model can be better specified, and which can take account of the versatility of the human brain.

Analysis of magnetoencephaoographic data as a nonlinear dynamical system

This thesis presents the results from an investigation into the merits of analysing Magnetoencephalographic (MEG) data in the context of dynamical systems theory. MEG is the study of both the methods for the measurement of minute magnetic flux variations at the scalp, resulting from neuro-electric activity in the neocortex, as well as the techniques required to process and extract useful information from these measurements. As a result of its unique mode of action - by directly measuring neuronal activity via the resulting magnetic field fluctuations - MEG possesses a number of useful qualities which could potentially make it a powerful addition to any brain researcher's arsenal. Unfortunately, MEG research has so far failed to fulfil its early promise, being hindered in its progress by a variety of factors. Conventionally, the analysis of MEG has been dominated by the search for activity in certain spectral bands - the so-called alpha, delta, beta, etc that are commonly referred to in both academic and lay publications. Other efforts have centred upon generating optimal fits of "equivalent current dipoles" that best explain the observed field distribution. Many of these approaches carry the implicit assumption that the dynamics which result in the observed time series are linear. This is despite a variety of reasons which suggest that nonlinearity might be present in MEG recordings. By using methods that allow for nonlinear dynamics, the research described in this thesis avoids these restrictive linearity assumptions. A crucial concept underpinning this project is the belief that MEG recordings are mere observations of the evolution of the true underlying state, which is unobservable and is assumed to reflect some abstract brain cognitive state. Further, we maintain that it is unreasonable to expect these processes to be adequately described in the traditional way: as a linear sum of a large number of frequency generators. One of the main objectives of this thesis will be to prove that much more effective and powerful analysis of MEG can be achieved if one were to assume the presence of both linear and nonlinear characteristics from the outset. Our position is that the combined action of a relatively small number of these generators, coupled with external and dynamic noise sources, is more than sufficient to account for the complexity observed in the MEG recordings. Another problem that has plagued MEG researchers is the extremely low signal to noise ratios that are obtained. As the magnetic flux variations resulting from actual cortical processes can be extremely minute, the measuring devices used in MEG are, necessarily, extremely sensitive. The unfortunate side-effect of this is that even commonplace phenomena such as the earth's geomagnetic field can easily swamp signals of interest. This problem is commonly addressed by averaging over a large number of recordings. However, this has a number of notable drawbacks. In particular, it is difficult to synchronise high frequency activity which might be of interest, and often these signals will be cancelled out by the averaging process. Other problems that have been encountered are high costs and low portability of state-of-the- art multichannel machines. The result of this is that the use of MEG has, hitherto, been restricted to large institutions which are able to afford the high costs associated with the procurement and maintenance of these machines. In this project, we seek to address these issues by working almost exclusively with single channel, unaveraged MEG data. We demonstrate the applicability of a variety of methods originating from the fields of signal processing, dynamical systems, information theory and neural networks, to the analysis of MEG data. It is noteworthy that while modern signal processing tools such as independent component analysis, topographic maps and latent variable modelling have enjoyed extensive success in a variety of research areas from financial time series modelling to the analysis of sun spot activity, their use in MEG analysis has thus far been extremely limited. It is hoped that this work will help to remedy this oversight.

Combining temporal and spectral information with spatial mapping to identify differences between phonological and semantic networks:a magnetoencephalographic approach

Early, lesion-based models of language processing suggested that semantic and phonological processes are associated with distinct temporal and parietal regions respectively, with frontal areas more indirectly involved. Contemporary spatial brain mapping techniques have not supported such clear-cut segregation, with strong evidence of activation in left temporal areas by both processes and disputed evidence of involvement of frontal areas in both processes. We suggest that combining spatial information with temporal and spectral data may allow a closer scrutiny of the differential involvement of closely overlapping cortical areas in language processing. Using beamforming techniques to analyze magnetoencephalography data, we localized the neuronal substrates underlying primed responses to nouns requiring either phonological or semantic processing, and examined the associated measures of time and frequency in those areas where activation was common to both tasks. Power changes in the beta (14-30 Hz) and gamma (30-50 Hz) frequency bandswere analyzed in pre-selected time windows of 350-550 and 500-700ms In left temporal regions, both tasks elicited power changes in the same time window (350-550 ms), but with different spectral characteristics, low beta (14-20 Hz) for the phonological task and high beta (20-30 Hz) for the semantic task. In frontal areas (BA10), both tasks elicited power changes in the gamma band (30-50 Hz), but in different time windows, 500-700ms for the phonological task and 350-550ms for the semantic task. In the left inferior parietal area (BA40), both tasks elicited changes in the 20-30 Hz beta frequency band but in different time windows, 350-550ms for the phonological task and 500-700ms for the semantic task. Our findings suggest that, where spatial measures may indicate overlapping areas of involvement, additional beamforming techniques can demonstrate differential activation in time and frequency domains. © 2012 McNab, Hillebrand, Swithenby and Rippon.

Ontology-based protein-protein interactions extraction from literature using the hidden vector state model

This paper proposes a novel framework of incorporating protein-protein interactions (PPI) ontology knowledge into PPI extraction from biomedical literature in order to address the emerging challenges of deep natural language understanding. It is built upon the existing work on relation extraction using the Hidden Vector State (HVS) model. The HVS model belongs to the category of statistical learning methods. It can be trained directly from un-annotated data in a constrained way whilst at the same time being able to capture the underlying named entity relationships. However, it is difficult to incorporate background knowledge or non-local information into the HVS model. This paper proposes to represent the HVS model as a conditionally trained undirected graphical model in which non-local features derived from PPI ontology through inference would be easily incorporated. The seamless fusion of ontology inference with statistical learning produces a new paradigm to information extraction.

A comparative study of bayesian models for unsupervised sentiment detection

This paper presents a comparative study of three closely related Bayesian models for unsupervised document level sentiment classification, namely, the latent sentiment model (LSM), the joint sentiment-topic (JST) model, and the Reverse-JST model. Extensive experiments have been conducted on two corpora, the movie review dataset and the multi-domain sentiment dataset. It has been found that while all the three models achieve either better or comparable performance on these two corpora when compared to the existing unsupervised sentiment classification approaches, both JST and Reverse-JST are able to extract sentiment-oriented topics. In addition, Reverse-JST always performs worse than JST suggesting that the JST model is more appropriate for joint sentiment topic detection.

A novel framework of training hidden Markov support vector machines from lightly-annotated data

Natural language understanding (NLU) aims to map sentences to their semantic mean representations. Statistical approaches to NLU normally require fully-annotated training data where each sentence is paired with its word-level semantic annotations. In this paper, we propose a novel learning framework which trains the Hidden Markov Support Vector Machines (HM-SVMs) without the use of expensive fully-annotated data. In particular, our learning approach takes as input a training set of sentences labeled with abstract semantic annotations encoding underlying embedded structural relations and automatically induces derivation rules that map sentences to their semantic meaning representations. The proposed approach has been tested on the DARPA Communicator Data and achieved 93.18% in F-measure, which outperforms the previously proposed approaches of training the hidden vector state model or conditional random fields from unaligned data, with a relative error reduction rate of 43.3% and 10.6% being achieved.

Scaling up question-answering to linked data

Linked Data semantic sources, in particular DBpedia, can be used to answer many user queries. PowerAqua is an open multi-ontology Question Answering (QA) system for the Semantic Web (SW). However, the emergence of Linked Data, characterized by its openness, heterogeneity and scale, introduces a new dimension to the Semantic Web scenario, in which exploiting the relevant information to extract answers for Natural Language (NL) user queries is a major challenge. In this paper we discuss the issues and lessons learned from our experience of integrating PowerAqua as a front-end for DBpedia and a subset of Linked Data sources. As such, we go one step beyond the state of the art on end-users interfaces for Linked Data by introducing mapping and fusion techniques needed to translate a user query by means of multiple sources. Our first informal experiments probe whether, in fact, it is feasible to obtain answers to user queries by composing information across semantic sources and Linked Data, even in its current form, where the strength of Linked Data is more a by-product of its size than its quality. We believe our experiences can be extrapolated to a variety of end-user applications that wish to scale, open up, exploit and re-use what possibly is the greatest wealth of data about everything in the history of Artificial Intelligence. © 2010 Springer-Verlag.

Cross ontology query answering on the semantic web:an initial evaluation

PowerAqua is a Question Answering system, which takes as input a natural language query and is able to return answers drawn from relevant semantic resources found anywhere on the Semantic Web. In this paper we provide two novel contributions: First, we detail a new component of the system, the Triple Similarity Service, which is able to match queries effectively to triples found in different ontologies on the Semantic Web. Second, we provide a first evaluation of the system, which in addition to providing data about PowerAqua's competence, also gives us important insights into the issues related to using the Semantic Web as the target answer set in Question Answering. In particular, we show that, despite the problems related to the noisy and incomplete conceptualizations, which can be found on the Semantic Web, good results can already be obtained.

Semantic search meets the Web

While semantic search technologies have been proven to work well in specific domains, they still have to confront two main challenges to scale up to the Web in its entirety. In this work we address this issue with a novel semantic search system that a) provides the user with the capability to query Semantic Web information using natural language, by means of an ontology-based Question Answering (QA) system [14] and b) complements the specific answers retrieved during the QA process with a ranked list of documents from the Web [3]. Our results show that ontology-based semantic search capabilities can be used to complement and enhance keyword search technologies.

AquaLog : an ontology-driven question answering system for organizational semantic intranets

The semantic web vision is one in which rich, ontology-based semantic markup will become widely available. The availability of semantic markup on the web opens the way to novel, sophisticated forms of question answering. AquaLog is a portable question-answering system which takes queries expressed in natural language and an ontology as input, and returns answers drawn from one or more knowledge bases (KBs). We say that AquaLog is portable because the configuration time required to customize the system for a particular ontology is negligible. AquaLog presents an elegant solution in which different strategies are combined together in a novel way. It makes use of the GATE NLP platform, string metric algorithms, WordNet and a novel ontology-based relation similarity service to make sense of user queries with respect to the target KB. Moreover it also includes a learning component, which ensures that the performance of the system improves over the time, in response to the particular community jargon used by end users.

AquaLog : an ontology-driven question answering system to interface the semantic web

The semantic web (SW) vision is one in which rich, ontology-based semantic markup will become widely available. The availability of semantic markup on the web opens the way to novel, sophisticated forms of question answering. AquaLog is a portable question-answering system which takes queries expressed in natural language (NL) and an ontology as input, and returns answers drawn from one or more knowledge bases (KB). AquaLog presents an elegant solution in which different strategies are combined together in a novel way. AquaLog novel ontology-based relation similarity service makes sense of user queries.

We know what you want to buy:a demographic-based system for product recommendation on microblogs

Product recommender systems are often deployed by e-commerce websites to improve user experience and increase sales. However, recommendation is limited by the product information hosted in those e-commerce sites and is only triggered when users are performing e-commerce activities. In this paper, we develop a novel product recommender system called METIS, a MErchanT Intelligence recommender System, which detects users' purchase intents from their microblogs in near real-time and makes product recommendation based on matching the users' demographic information extracted from their public profiles with product demographics learned from microblogs and online reviews. METIS distinguishes itself from traditional product recommender systems in the following aspects: 1) METIS was developed based on a microblogging service platform. As such, it is not limited by the information available in any specific e-commerce website. In addition, METIS is able to track users' purchase intents in near real-time and make recommendations accordingly. 2) In METIS, product recommendation is framed as a learning to rank problem. Users' characteristics extracted from their public profiles in microblogs and products' demographics learned from both online product reviews and microblogs are fed into learning to rank algorithms for product recommendation. We have evaluated our system in a large dataset crawled from Sina Weibo. The experimental results have verified the feasibility and effectiveness of our system. We have also made a demo version of our system publicly available and have implemented a live system which allows registered users to receive recommendations in real time. © 2014 ACM.