A5062879885- Topic Modeling
- Speech and dialogue systems
- Natural Language Processing Techniques
- Speech Recognition and Synthesis
- Sentiment Analysis and Opinion Mining
- Emotion and Mood Recognition
- Multi-Agent Systems and Negotiation
- Advanced Text Analysis Techniques
- EEG and Brain-Computer Interfaces
- Mental Health via Writing
- Algorithms and Data Compression
- Machine Learning and Algorithms
- Digital Mental Health Interventions
- Language, Metaphor, and Cognition
- Humor Studies and Applications
- Personality Traits and Psychology
- AI in Service Interactions
- Text Readability and Simplification
- ECG Monitoring and Analysis
- Language, Discourse, Communication Strategies
- Mental Health Research Topics
- Music and Audio Processing
- Authorship Attribution and Profiling
- Text and Document Classification Technologies
- Semantic Web and Ontologies
AT&T (United States)
1999-2023
University of Trento
2014-2023
Philips (Netherlands)
2021
University of Strathclyde
2019
University of the Basque Country
2005
Nokia (United States)
2002
Spoken Language Understanding (SLU) for conversational systems (SDS) aims at extracting concept and their relations from spontaneous speech. Previous approaches to SLU have modeled as stochastic semantic networks ranging generative approach discriminative. As spoken dialog complexity increases, needs perform understanding based on a richer set of features a-priori knowledge, long dependency, history, system belief, etc. This paper studies discriminative modeling the sentence segmentation...
We are interested in the problem of adaptive learning context automatic speech recognition (ASR). In this paper, we propose an active algorithm for ASR. Automatic systems trained using human supervision to provide transcriptions utterances. The goal Active Learning is minimize training acoustic and language models maximize performance given transcribed untranscribed data. aims at reducing number examples be labeled by automatically processing unlabeled examples, then selecting most...
Semantics deals with the organization of meanings and relations between sensory signs or symbols what they denote mean. Computational semantics performs a conceptualization world using computational processes for composing meaning representation structure from available their features present, example, in words sentences. Spoken language understanding (SLU) is interpretation conveyed by speech signal. SLU natural (NLU) share goal obtaining conceptual Specific to fact that be used are coded...
For the natural and social interaction it is necessary to understand human behavior. Personality one of fundamental aspects, by which we can behavioral dispositions. It evident that there a strong correlation between users’ personality way they behave on online network (e.g., Facebook). This paper presents automatic recognition Big-5 traits (Facebook) using status text. studied different classification methods such as SMO (Sequential Minimal Optimization for Support Vector Machine), Bayesian...
While mental health applications are increasingly becoming available for large populations of users, there is a lack controlled trials on the impacts such applications. Artificial intelligence (AI)-empowered agents have been evaluated when assisting adults with cognitive impairments; however, few aging who still actively working. These often high stress levels related to changes in their work places, and symptoms eventually affect quality life.We aimed evaluate contribution TEO (Therapy...
We address the problem of computing a consensus translation given outputs from set machine (MT) systems. The translations MT systems are aligned with multiple string alignment algorithm and is then computed. describe hypothesis computation. report on subjective objective performance multilingual acquisition approach limited domain spoken language application. evaluate five domain-independent off-the-shelf show that consensus-based equal to or better than any systems, in terms both measures.
This paper explores the use of innovative kernels based on syntactic and semantic structures for a target relation extraction task. Syntax is derived from constituent dependency parse trees whereas semantics concerns to entity types lexical sequences. We investigate effectiveness such representations in automated texts. process above data by means Support Vector Machines along with tree, partial tree word sequence kernels. Our study ACE 2004 corpus illustrates that combination achieves high...
One of the first steps in building a spoken language understanding (SLU) module for dialogue systems is extraction flat concepts out given word sequence, usually provided by an automatic speech recognition (ASR) system. In this paper, six different modeling approaches are investigated to tackle task concept tagging. These methods include classical, well-known generative and discriminative like Finite State Transducers (FSTs), Statistical Machine Translation (SMT), Maximum Entropy Markov...
The Workshop on Computational Personality Recognition aims to define the state-of-the-art in field and provide tools for future standard evaluations personality recognition tasks. In WCPR14 we released two different datasets: one of Youtube Vlogs Mobile Phone interactions. We structured workshop tracks: an open shared task, where participants can do any kind experiment, a competition. also distinguished tasks: A) from multimedia data, B) text only. this paper discuss results workshop.
Measuring personality traits has a long story in psychology where analysis been done by asking sets of questions. These question (inventories) have designed investigating lexical terms that we use our daily communications or analyzing biological phenomena. Whether consciously unconsciously express thoughts and behaviors when communicating with others, either verbally, non-verbally using visual expressions. Recently, research behavioral signal processing focused on automatically measuring...
State-of-the-art speech recognition systems are trained using transcribed utterances, preparation of which is labor intensive and time-consuming. In this paper, we describe a new method for reducing the transcription effort training in automatic (ASR). Active learning aims at number examples to be labeled by automatically processing unlabeled examples, then selecting most informative ones with respect given cost function human label. We estimate confidence score each word utterance,...
Spoken language understanding (SLU) aims at extracting meaning from natural speech. Over the past decade, a variety of practical goal-oriented spoken dialog systems have been built for limited domains. SLU in these ranges predetermined phrases through fixed grammars, some predefined named entities, users' intents call classification, to combinations and entities. In this paper, we present system VoiceTone/spl reg/ (a service provided by AT&T where develops, deploys hosts applications...
Most research that explores the emotional state of users spoken dialog systems does not fully utilize contextual nature structure provides. This paper reports results machine learning experiments designed to automatically classify user turns using a corpus 5,690 dialogs collected with “How May I Help You SM ” system. We show augmenting standard lexical and prosodic features exploit track stateincreases classification accuracy by 2.6%.
LLMs' sources of knowledge are data snapshots containing factual information about entities collected at different timestamps and from media types (e.g. wikis, social media, etc.). Such unstructured is subject to change due updates through time past present. Equally important the inconsistencies inaccuracies occurring in sources. Consequently, model's an entity may be perturbed while training over sequence or inference time, resulting inconsistent inaccurate model performance. In this work,...
This paper describes the AT&T WATSON real-time speech recognizer, product of several decades research at AT&T. The recognizer handles a wide range vocabulary sizes and is based on continuous-density hidden Markov models for acoustic modeling finite state networks language modeling. recognition network optimized efficient search. We identify algorithms used high-accuracy, low-latency recognition. present results small large tasks taken from VoiceTone/sup /spl reg// service, showing word...