Individuals with intellectual disability often face challenges in expressing intentions and initiating conversations. While assistive communication technologies typically emphasize language acquisition, generic image browsers enable open-ended self-expression without structured symbols or language. Building on prior work showing that intent-based image recommendations represent potential meanings of a selected image, we examine how such suggestions are taken up in real communicative practice. We studied an assistive communication prototype that uses generative AI to provide intent-based image suggestions. The study was conducted across multiple sessions with 15 adults with intellectual disability, varying facilitation and interaction framing to examine how these suggestions shaped conversational flow. Our findings indicate that intent-based image suggestions played multiple interactional roles, including prompting expression, clarification, and sustaining conversation. We also identify moments of misalignment, where suggestions failed to align with users’ communicative goals due to missing cultural context or interactional support. We discuss interaction mechanisms and design implications for intent-based assistive communication systems to support inclusive and effective communication.

Digital assistive technologies (ATs) have been widely used to support people with disabilities at work.However, many existing systems, interfaces, and tools remain inaccessible or insufficiently adaptable to the wide range of human abilities, particularly when cognitive, communicative, or sensory differences are involved.This gap is further exacerbated by what scholars and activists refer to as the disability divide: the sociotechnical disparity between people with and without disabilities in terms of access to, use of, and benefits from digital technologies.Despite increasing policy efforts and legal frameworks, vocational inclusion and training remains a significant challenge.By bringing together a diverse community, this workshop seeks to critically examine the role of digital ATs in advancing vocational inclusion for individuals with disabilities. CCS Concepts• Human-

This experience report offers the perspective of the first author, Chloe, a woman with intellectual disability, on participation in technology co-design and her views of how technology, particularly social robots, can be designed for everyone’s benefit. The core of the report are quotes from Chloe in response to questions of the research team, with some context in terms of literature and background for her experiences. A short discussion contributed by the academic researcher authors reflects on relationship building throughout series of co-design research projects and highlights future directions for inclusive co-design research, the nuances between co-design research participants and co-researchers.

Advanced sensing capabilities are emerging in smart conversational systems. Without support, keyword-based searching can be challenging, requiring expertise, experience and the ability to formulate abstract knowledge of the information-seeking process. However, people with intellectual disability are often comfortable communicating in conversations that allow for both verbal and visual communication. This paper presents an understanding of multimodal and collaborative search systems for people with intellectual disability. First, we present a Wizard of Oz probe, called MMCS, that provides flexibility and customisation to explore different modalities and opportunities for accessible opportunities. We present the outcomes of an ethnographic study conducted in a collaborative setting with twenty participants across four sessions and follow-up interviews. We found that multimodal and conversational interaction can play a crucial role in social support, peer awareness, and personal interests. Finally, we provide design implications and future research directions towards understanding, responding, and reformulating query intent for multimodal and collaborative search systems.

Non-verbal communication is essential for the social inclusion of individuals with an intellectual disability, affecting interactions with others as well as technological systems. This study focuses on non-symbolic communication of people with intellectual disability through generic images without specific or detailed subject matter. A key challenge in this medium is discerning the underlying intentions behind images selected as visual prompts for conversation.

AI-driven commercial innovations and the digital disruptions they create, tend to accelerate faster than assistive technologies, and are rarely designed with inclusion and diversity in mind. We explore the joint value of research through design and co-design to give a voice to users with intellectual disability to set new directions for inclusive innovation. To do this, we present an account of, and a reflection on, the reframing that took place throughout a research program that has evolved over the last 8 years, presented through the lens of 3 case studies. These illustrate turning points in the frames of the research and its journey through the disciplinary traditions of Information Retrieval and Human Computer Interaction (HCI). The contributions of this paper are threefold. First, we contribute knowledge on the value of research through design to identify new frames for inclusive intelligent systems. Second, we extend inclusive co-design approaches to employing working prototypes that can support participant’s voice about the design of the algorithms that underpin intelligent systems. We highlight how these working prototypes nurture the importance of participation and observation. Third, we contribute new frames for inclusive information retrieval, with new perspectives on intent, particularly in the context of image search.

People with intellectual disabilities often experience inequalities that affect the standard of their everyday lives. Assistive technologies can help alleviate some of these inequalities, yet abandonment rates remain high. This is in part due to a lack of involvement of all stakeholders in their design and evaluation, thus resulting in outputs that do not meet this cohort’s complex and heterogeneous needs. The aim of this half-day workshop is to focus on community building in a field that is relatively thin and disjointed, thereby enabling researchers to share experiences on how to design for and with people with intellectual disabilities, provide internal support, and establish new collaborations. Workshop outcomes will help to fill a gap in the available guidelines on how to include people with intellectual disabilities in research, through more accessible protocols as well as personalised and better fit-for-purpose technologies.

This research aims to explore how tangible technology created through co-design can be designed in a way that invites social interaction for people with intellectual disability. We conducted co-design sessions with one participant to create a sensory musical blanket. As the trials were run in a collective environment, their peers were drawn to the developing design. The design method and unique interactions are key contributions of this research.

This paper aims to create a tangible design framework for practitioners to follow when designing an online social media platform for individuals with intellectual disability. Currently, legislation and best practice consider cyber security and safety for the general public, giving particular attention to the protection of children. However, despite the support in health care, financial assistance, and education, individuals with intellectual disability are rarely considered when it comes to cybersafety. To achieve inclusivity, an integrative review was conducted to make connections between disciplines of education and information technology and law. The process was split into three phases: (i) understanding the challenges those with intellectual disability face, both when using a social media interface and when evaluating safety risks; (ii) identifying gaps and understanding the implications for persons with intellectual disability from legislative and design and design principles; and (iii) visualisation of data flow to model interactions. In conclusion, an inclusive framework is proposed for practitioners when designing online social media platforms for people with intellectual disability.

This research explores the conditions and opportunities for citizen science applications to enhance their accessibility to people with intellectual disability (ID). In this paper, we present how the knowledge gathered by co-designing with a group of 3 participants with ID led to a design judged accessible and engaging by another group of 4 participants with ID. We contribute the key elements of that design: static subject, visual engagement, embodiment and social connectedness.

This paper presents the first iteration of the design of a web application which supports its users to access and arrange pictures as a non-linguistic way of supporting communication. We motivate our initial design by examining related work on Augmentative Alternative Communication (AAC). We present our reflections on the use of a working prototype by two minimally-verbal users with intellectual disability and how this can inform future work.

In recent work, design researchers have sought to ensure that people with disabilities are engaged as competent and valued contributors to co-design. Yet, little is known about how to achieve this with adults with severe intellectual disabilities. Navigating design in the context of complex care practices is challenging, charged with uncertainty, and requires sustained effort of methodological and affective adjustments. To establish a respectful co-design relationship and enrich participation, we turn to Active Support (AS), an evidence-based strategy for engaging adults with severe intellectual disabilities. We present a reflective account of long-term field work that utilized the four aspects of AS, a) every moment has potential; b) graded assistance; c) little and often; d) maximizing choice and control. We discuss how these principles contribute to deepening HCI methods by ensuring interactional turns for adults with severe disabilities, revealing their unique competences, thereby shaping design direction and providing design insight.

People with intellectual disability (ID) deserve appropriate access to information online. A vast amount of information on the Internet is written text in the form of articles, and it is often said that summarising these texts could enhance their accessibility. This qualitative research investigates how people with ID prefer to gather information from articles, either in their original form or automatically summarised. The researchers observed the choices and strategies of 10 participants with ID through the reading process, and conducted contextual interviews to understand their preferences, attitudes and the difficulties they faced. The study found that the length of the article is only secondary to the relevance of the article to the person's interests. Summarised articles were found easier to read and having familiar words and images to supplement the text can help people with intellectual disability understand what an article is about.

In this paper, we present a case study of the iterative design of TalkingBox, a communication device designed with a person with a severe cognitive disability and his support network. TalkingBox combines graphic symbols with tangible technology to foster the use of symbolic communication by leveraging the person's strength and interest in memory matching games. In the course of designing, trialing and iterating the TalkingBox, we discovered that the design supported not only the development of symbolic communication, but also revealed new interests and strengths of our participant. TalkingBox highlighted opportunities for interactions with peers, revealed new skills in visual discrimination, and evidenced interests. These could, in turn, support staff and family to adapt their support. More importantly, TalkingBox had become a living portfolio presenting our participant with severe disability through the lens of their strengths. We discuss opportunities for research through co-design to open new avenues for future communication technologies.

Expressing one's thoughts and feelings is a fundamental human need - the basis for communication and social interaction. We ask, how do minimally-verbal children on the autism spectrum express themselves? How can we better recognise instances of self-expression? And how might technologies support and encourage self-expression? To address these questions, we undertook co-design research at an autism-specific primary school with 20 children over one school year. This paper contributes six Modalities of Self-Expression, through which children self-express and convey their design insights. Each modality of self-expression can occur across two different dimensions (socio-expressive and auto-expressive) and can be of a fundamental or an integrative nature. Further, we contribute the design trajectory of a tangible ball prototype, the ExpressiBall, which - through voice, sounds, lights, and motion sensors - explores how tangible technologies can support this range of expressive modalities. Finally, we discuss the concept of Self-Expression by Design.

Traditional methods of engagement with pre-defined queries, verbal instruction and interviewing do not provide necessary means to address information-seeking behavior and visual browsing for participants with severe autism and intellectual disability. In this paper, we identify challenges and characteristics of providing effective methods to explore visual browsing and video recommender systems with one non-verbal participant with autism and intellectual disability. We contribute a case study and a reflection on a) how iterative design approaches that builds on special interests and strengths of one individual with disability can support experimental IIR research in becoming more inclusive, b) some of the ethical consideration that arise in the tensions between participation in the research and other interests and c) how flexible experimental and apparatus design can further allow participant's terms to prevail.

People of all abilities deserve to be included in the research and design of information access technologies. In this tutorial, we will explore the latest frameworks and approaches in interaction design that seek to engage and design with people with diverse cognitive abilities. Frameworks on ability-based design and respectful design suggest starting from users abilities, personalising, and engaging in co-design activities in order to truly understand users. Frameworks on design-after design and co-design beyond words provide tools and approaches to engage with users with intellectual disability, who may not be verbal, in co-design exercises. This tutorial informs participants about various frameworks and ways of enacting them in a participatory and hands-on approach.

At CHIIR 2019, Berget and MacFarlane [4] pointed out the need for ethical methodologies when involving participants with dyslexia. In this paper, we further propose that a stance of ability based design and participatory design approaches can further involve, engage and support people with intellectual disability in interactive information retrieval (IIR) research. Through a case study with an accessible prototype designed to access instructional videos, we demonstrate how an approach building on participant's interests and providing them support as part of the study design leads to ecologically valid observations. The accessible prototype makes use of images as prompts and query support, and includes social aspects. Our observations confirm that users with intellectual disability favour a visual approach to information access and interaction. The contributions of this work are primarily 1) a 2 step approach with supported participatory design approaches involving early prototypes 2) a case study of this approach to investigate information access interfaces with people with intellectual disability and 3) a reflection on the case study and applicability of the method in IIR evaluation.

Online videos are a medium of choice for young adults to access or receive information, and recent work has highlighted that it is a particularly effective medium for adults with intellectual disability, by its visual nature. Reflecting on a case study presenting fieldwork observations of how adults with intellectual disability engage with videos on the Youtube platform, we propose a framework to define and evaluate the accessibility of such large video repositories, from an informational perspective. The proposed framework nuances the concept of information accessibility from that of the accessibility of information access interfaces themselves (generally catered for under web accessibility guidelines), or that of the documents (generally covered in general accessibility guidelines). It also includes a notion of search (or browsing) accessibility, which reflects the ability to reach the document containing the information. In the context of large information repositories, this concept goes beyond how the documents are organized into how automated processes (browsing or searching) can support users. In addition to the framework we also detail specifics of document accessibility for videos. The framework suggests a multi-dimensional approach to information accessibility evaluation which includes both cognitive and sensory aspects. This framework can serve as a basis for practitioners when designing video information repositories accessible to people with intellectual disability, and extends on the information presentation guidelines such as suggested by the WCAG.

Young adults with intellectual disability are keen users of social media. However, there is little understanding about how their skills and participation in social media might be leveraged to support further skills development. Employing a participatory approach through workshops with eleven participants and interviews with eight parents, we investigated what skills young adults desire and how they might be able to leverage their participation in YouTube and Facebook to develop these skills. We found that young adults want to develop social skills of such as playing sports or learning languages, but that leveraging social media participation to do this goes beyond their typical use, and requires both collaborative support and accessible design. Based on these findings, we propose and discuss a collaboration-in-the-loop framework that integrates support through personal networks and an accessible user interface design. We conclude with a reflection on designing to leverage participation, interests and competencies to support people with intellectual disability.

As head mounted displays and 360° video cameras are becoming affordable, they offer opportunities to personalise immersive learning experiences to local contexts and individuals. In this paper, we present lessons learnt from a participatory design process focused on understanding engagement and preferences of users with intellectual disability viewing 360° videos. Over 4 iterations involving re-designs informed by interviews and observations of two to four participants with intellectual disability, we have established that: participants are more comfortable with using the technology if they are first introduced to a familiar scene before seeing anything new, they prefer to be 'accompanied' by an in-video facilitator, and participants engage more with the immersive visual environment when prompted to look around from within the video. We have also established a number of guidelines for filming with a 360° camera with regards to movement and viewpoint.

Existing co-design methods support verbal children on the autism spectrum in the design process, while their minimally-verbal peers are overlooked. We describe Co-Design Beyond Words (CDBW), an approach which merges existing co-design methods with practice-based methods from Speech and Language Therapy which are child-led and interests-based. These emphasise the rich detail that can be conveyed in the moment, through recognising occurrences of, for example, Joint Attention, Turn Taking and Imitation. We worked in an autism-specific primary school over 20 weeks with ten children, aged 5 to 8. We co-designed a playful prototype, the TangiBall, using the three iterative phases of CDBW; the Foundation Phase (preparation for interaction), the Interaction Phase (designing-and-reflecting in the moment) and the Reflection Phase (reflection-on-action). We contribute a novel co-design approach and present moments of interaction, the micro instances in design in which minimally-verbal children on the spectrum can convey meaning beyond words, through their actions, interactions, and attentional foci. These moments of interaction provide design insight, shape design direction, and reveal unique strengths, interests, and abilities.

A person with a disability has to assemble support services and technologies from different organisations in order to live well, which may require help from family. We call this assembling of services and technologies personal infrastructuring, the process of learning about how to navigate the world, what support is available, and how to obtain and design new support through various organisational infrastructures. Such infrastructures include disability services organisations, the health sector, community organisations, and friend and family networks. Our vision was to explore how a person with a disability might engage in design with volunteer designers to meet their unique needs that were not met by their existing infrastructure of organisations, products and services. Through codesign with two people and their families, we developed design artefacts such as user profiles and video stories to support communication, mutual learning, need finding and need expression. We discovered that these design artefacts were used beyond their immediate purposes of design to further support their personal infrastructuring. In this paper, we discuss how understandings of infrastructure and infrastructuring from Science and Technology Studies and Information Systems translate into familial contexts and the concept of personal infrastructuring.

Digital technologies to support children on the autism spectrum often offer predefined content for modelling, communicating and training. However, children may not relate to the content, and it may not match their own personal interests and motivations. This paper investigates the use of MyWord, an interest-based, child-led technology, as an exploratory probe. This audio-visual dictionary app supports a child to build their own personalised catalogue of favourite words, images and audio over time. We undertook a field study over two school terms in an autism-specific primary school with 12 minimally-verbal children aged 5 to 8 and their teachers and speech therapists. Findings indicate that creating dictionary entries involved processes of personal choice, representation of the self and interests, and dynamic action and play. Use of personally and contextually relevant words enhanced engagement, interaction and self-expression. We contribute a novel, flexible, interest-based technology, and reflections on its use in autism-specific school contexts. We highlight the importance of the child's lived experience and holistic child-led approaches to technology design.

Technologies to support children with autism tend to use predefined content to enhance specific skills, such as verbal communication or emotion recognition. Few mobilise the child's own (often very specific) interests, strengths and capabilities. Digital technologies offer opportunities for children to personalise learning with their own content, following their own interests and enabling their self-expression. This project sought to engage children to record and express their own interests within their contexts of support - the home and the classroom. The vehicle for self-expression was an audio-visual calendaring app called MeCalendar. Implementation was kept open-ended to allow teachers to use it in ways that best fit with their existing embedded practices. In this paper we report on how the prototype has been appropriated in two classrooms by teachers in an autism-specific school setting with children aged 6 to 7. Our contribution is an understanding of how technologies for self-expression led to enhanced verbal communication, positive reinforcement through video modelling, engagement in class tasks and enhanced social interaction. Children appropriated the design in unimagined ways, leading them to self-scaffold and to catalyse their confidence in social interaction and self-expression. Teachers played an integral role in appropriating the design in the classroom, specifically through their in-depth knowledge of each child and their individual needs, strengths and interests.

This article demonstrates the benefits of using sequence representations based on word embeddings to inform the seed selection and sample selection processes in an active learning pipeline for clinical information extraction. Seed selection refers to choosing an initial sample set to label to form an initial learning model. Sample selection refers to selecting informative samples to update the model at each iteration of the active learning process. Compared to supervised machine learning approaches, active learning offers the opportunity to build statistical classifiers with a reduced amount of training samples that require manual annotation. Reducing the manual annotation effort can support automating the clinical information extraction process. This is particularly beneficial in the clinical domain, where manual annotation is a time‐consuming and costly task, as it requires extensive labor from clinical experts. Our empirical findings demonstrate that (a) using sequence representations along with the length of sequence for seed selection shows potential towards more effective initial models, and (b) using sequence representations for sample selection leads to significantly lower manual annotation efforts, with up to 3% and 6% fewer tokens and concepts requiring annotation, respectively, compared to state‐of‐the‐art query strategies.

Topic Modeling algorithms are rarely used to support the qualitative content analysis process. The main contributing factors for the lack of mainstream adoption can be attributed to the perception that Topic Modeling produces topics of poor quality and that content analysts do not trust the derived topics because they are unable to supply domain knowledge and interact with the algorithm. In this paper, interactive Topic Modeling algorithms namely Dirichlet-Forrest Latent Dirichlet Allocation and Penalised Non-negative Matrix Factorisation, are evaluated with respect to their ability to aid qualitative content analysis. More specifically, the relationship between interactivity, interpretation, topic coherence and trust in interactive content analysis is examined. The findings indicate that providing content analysts with the ability to interact with Topic Modeling algorithms produces topics that are directly related to their research questions. However, a number of improvements to these algorithms were also identified which have the potential to influence future algorithm development to better meet the requirements of qualitative content analysts.

OBJECTIVE: This paper presents an automatic, active learning-based system for the extraction of medical concepts from clinical free-text reports. Specifically, (1) the contribution of active learning in reducing the annotation effort and (2) the robustness of incremental active learning framework across different selection criteria and data sets are determined. MATERIALS AND METHODS: The comparative performance of an active learning framework and a fully supervised approach were investigated to study how active learning reduces the annotation effort while achieving the same effectiveness as a supervised approach. Conditional random fields as the supervised method, and least confidence and information density as 2 selection criteria for active learning framework were used. The effect of incremental learning vs standard learning on the robustness of the models within the active learning framework with different selection criteria was also investigated. The following 2 clinical data sets were used for evaluation: the Informatics for Integrating Biology and the Bedside/Veteran Affairs (i2b2/VA) 2010 natural language processing challenge and the Shared Annotated Resources/Conference and Labs of the Evaluation Forum (ShARe/CLEF) 2013 eHealth Evaluation Lab. RESULTS: The annotation effort saved by active learning to achieve the same effectiveness as supervised learning is up to 77%, 57%, and 46% of the total number of sequences, tokens, and concepts, respectively. Compared with the random sampling baseline, the saving is at least doubled. CONCLUSION: Incremental active learning is a promising approach for building effective and robust medical concept extraction models while significantly reducing the burden of manual annotation.

This paper presents a new active learning query strategy for information extraction, called Domain Knowledge Informativeness (DKI). Active learning is often used to reduce the amount of annotation effort required to obtain training data for machine learning algorithms. A key component of an active learning approach is the query strategy, which is used to iteratively select samples for annotation. Knowledge resources have been used in information extraction as a means to derive additional features for sample representation. DKI is, however, the first query strategy that exploits such resources to inform sample selection. To evaluate the merits of DKI, in particular with respect to the reduction in annotation effort that the new query strategy allows to achieve, we conduct a comprehensive empirical comparison of active learning query strategies for information extraction within the clinical domain. The clinical domain was chosen for this work because of the availability of extensive structured knowledge resources which have often been exploited for feature generation. In addition, the clinical domain offers a compelling use case for active learning because of the necessary high costs and hurdles associated with obtaining annotations in this domain. Our experimental findings demonstrated that (1) amongst existing query strategies, the ones based on the classification model's confidence are a better choice for clinical data as they perform equally well with a much lighter computational load, and (2) significant reductions in annotation effort are achievable by exploiting knowledge resources within active learning query strategies, with up to 14% less tokens and concepts to manually annotate than with state-of-the-art query strategies.

Advances in neural network language models have demonstrated that these models can effectively learn representations of words meaning. In this paper, we explore a variation of neural language models that can learn on concepts taken from structured ontologies and extracted from free-text, rather than directly from terms in free-text.

A user's query is considered to be an imprecise description of their information need. Automatic query expansion is the process of reformulating the original query with the goal of improving retrieval effectiveness. Many successful query expansion techniques model syntagmatic associations that infer two terms co‐occur more often than by chance in natural language. However, structural linguistics relies on both syntagmatic and paradigmatic associations to deduce the meaning of a word. Given the success of dependency‐based approaches to query expansion and the reliance on word meanings in the query formulation process, we argue that modeling both syntagmatic and paradigmatic information in the query expansion process improves retrieval effectiveness. This article develops and evaluates a new query expansion technique that is based on a formal, corpus‐based model of word meaning that models syntagmatic and paradigmatic associations. We demonstrate that when sufficient statistical information exists, as in the case of longer queries, including paradigmatic information alone provides significant improvements in retrieval effectiveness across a wide variety of data sets. More generally, when our new query expansion approach is applied to large‐scale web retrieval it demonstrates significant improvements in retrieval effectiveness over a strong baseline system, based on a commercial search engine.

We identify Relation Completion (RC) as one recurring problem that is central to the success of novel big data applications such as Entity Reconstruction and Data Enrichment.Given a semantic relation R, RC attempts at linking entity pairs between two entity lists under the relation R. To accomplish the RC goals, we propose to formulate search queries for each query entity α based on some auxiliary information, so that to detect its target entity β from the set of retrieved documents.For instance, a Pattern-based method (PaRE) uses extracted patterns as the auxiliary information in formulating search queries.However, high-quality patterns may decrease the probability of finding suitable target entities.As an alternative, we propose CoRE method that uses context terms learned surrounding the expression of a relation as the auxiliary information in formulating queries.The experimental results based on several real-world web data collections demonstrate that CoRE reaches a much higher accuracy than PaRE for the purpose of RC.

Many successful query expansion techniques ignore information about the term dependencies that exist within natural language. However, researchers have recently demonstrated that consistent and significant improvements in retrieval effectiveness can be achieved by explicitly modelling term dependencies within the query expansion process. This has created an increased interest in dependency-based models.

In this paper, we propose a new type of Dictionary-based Entity Recognition Problem, named Approximate Membership Localization (AML). The popular Approximate Membership Extraction (AME) provides a full coverage to the true matched substrings from a given document, but many redundancies cause a low efficiency of the AME process and deteriorate the performance of real-world applications using the extracted substrings. The AML problem targets at locating nonoverlapped substrings which is a better approximation to the true matched substrings without generating overlapped redundancies. In order to perform AML efficiently, we propose the optimized algorithm P-Prune that prunes a large part of overlapped redundant matched substrings before generating them. Our study using several real-word data sets demonstrates the efficiency of P-Prune over a baseline method. We also study the AML in application to a proposed web-based join framework scenario which is a search-based approach joining two tables using dictionary-based entity recognition from web documents. The results not only prove the advantage of AML over AME, but also demonstrate the effectiveness of our search-based approach.

Measures of semantic similarity between medical concepts are central to a number of techniques in medical informatics, including query expansion in medical information retrieval. Previous work has mainly considered thesaurus-based path measures of semantic similarity and has not compared different corpus-driven approaches in depth. We evaluate the effectiveness of eight common corpus-driven measures in capturing semantic relatedness and compare these against human judged concept pairs assessed by medical professionals. Our results show that certain corpus-driven measures correlate strongly (approx 0.8) with human judgements. An important finding is that performance was significantly affected by the choice of corpus used in priming the measure, i.e., used as evidence from which corpus-driven similarities are drawn. This paper provides guidelines for the implementation of semantic similarity measures for medical informatics and concludes with implications for medical information retrieval.

This paper develops and evaluates an enhanced corpus based approach for semantic processing. Corpus based models that build representations of words directly from text do not require pre-existing linguistic knowledge, and have demonstrated psychologically relevant performance on a number of cognitive tasks. However, they have been criticised in the past for not incorporating sufficient structural information. Using ideas underpinning recent attempts to overcome this weakness, we develop an enhanced tensor encoding model to build representations of word meaning for semantic processing. Our enhanced model demonstrates superior performance when compared to a robust baseline model on a number of semantic processing tasks.

In a pilot application based on web search engine called Web-based Relation Completion (WebRC), we propose to join two columns of entities linked by a predefined relation by mining knowledge from the web through a web search engine. To achieve this, a novel retrieval task Relation Query Expansion (RelQE) is modelled: given an entity (query), the task is to retrieve documents containing entities in predefined relation to the given one. Solving this problem entails expanding the query before submitting it to a web search engine to ensure that mostly documents containing the linked entity are returned in the top K search results. In this paper, we propose a novel Learning-based Relevance Feedback (LRF) approach to solve this retrieval task. Expansion terms are learned from training pairs of entities linked by the predefined relation and applied to new entity-queries to find entities linked by the same relation. After describing the approach, we present experimental results on real-world web data collections, which show that the LRF approach always improves the precision of top-ranked search results to up to 8.6 times the baseline. Using LRF, WebRC also shows performances way above the baseline.

This paper presents a framework for evaluating information retrieval of medical records. We use the BLULab corpus, a large collection of real-world de-identified medical records. The collection has been hand coded by clinical terminol- ogists using the ICD-9 medical classification system. The ICD codes are used to devise queries and relevance judge- ments for this collection. Results of initial test runs using a baseline IR system are provided. Queries and relevance judgements are online to aid further research in medical IR. Please visit: http://koopman.id.au/med_eval.

In this paper, we propose a search-based approach to join two tables in the absence of clean join attributes. Non-structured documents from the web are used to express the correlations between a given query and a reference list. To implement this approach, a major challenge we meet is how to efficiently determine the number of times and the locations of each clean reference from the reference list that is approximately mentioned in the retrieved documents. We formalize the Approximate Membership Localization (AML) problem and propose an efficient partial pruning algorithm to solve it. A study using real-word data sets demonstrates the effectiveness of our search-based approach, and the efficiency of our AML algorithm.

This paper introduces a method combining spell checking and phonetic interpretation in order to automatically rewrite questions typed by dyslexic spellers. The method uses a finite state automata framework. Dysorthographics refers to incorrect word segmentation which usually causes classical spelling correc-tors fail. The specificities of the information retrieval context are that flexion errors have no impact since the sentences are lemmatised and filtered and that several hypothesis can be processed for one query. Our system is evaluated on questions collected with the help of an orthophonist. The word error rate on lemmatised sentences falls from 60% to 22% (falls to 0% on 43% of sentences).

International audience