Multiple Corpora Research Articles

PUnifiedNER: A Prompting-Based Unified NER System for Diverse Datasets

Much of named entity recognition (NER) research focuses on developing dataset-specific models based on data from the domain of interest, and a limited set of related entity types. This is frustrating as each new dataset requires a new model to be trained and stored. In this work, we present a ``versatile'' model---the Prompting-based Unified NER system (PUnifiedNER)---that works with data from different domains and can recognise up to 37 entity types simultaneously, and theoretically it could be as many as possible. By using prompt learning, PUnifiedNER is a novel approach that is able to jointly train across multiple corpora, implementing intelligent on-demand entity recognition. Experimental results show that PUnifiedNER leads to significant prediction benefits compared to dataset-specific models with impressively reduced model deployment costs. Furthermore, the performance of PUnifiedNER can achieve competitive or even better performance than state-of-the-art domain-specific methods for some datasets. We also perform comprehensive pilot and ablation studies to support in-depth analysis of each component in PUnifiedNER.

Speech emotion recognition based on syllable-level feature extraction

Speech emotion recognition systems have high computational requirements for deep learning models and low generalizability mainly because of the poor reliability of emotional measurements across multiple corpora. To solve these problems, we present a speech emotion recognition system based on a reductionist approach of decomposing and analyzing syllable-level features. Mel-spectrogram of an audio stream is decomposed into syllable-level components, which are then analyzed to extract statistical features. The proposed method uses formant attention, noise-gate filtering, and rolling normalization contexts to decrease contextual differences and increase focus the attention on the structure of a formant. A set of syllable-level formant features is extracted and fed into a single hidden layer neural network that makes predictions for each syllable as opposed to the conventional approach of using a sophisticated deep learner to make sentence-wide predictions. The syllable level predictions help to lower the aggregated error in utterance level cross-corpus predictions. The experiments on IEMOCAP (IE), MSP-Improv (MI), and RAVDESS (RA) databases show that the method archives better than the state-of-the-art cross-corpus unweighted accuracy of 47.6% for IE to MI and 56.2% for MI to IE.

Measuring the interdisciplinary characteristics of Chinese research in library and information science based on knowledge elements

PurposeThis study analyzed the interdisciplinary characteristics of Chinese research studies in library and information science (LIS) measured by knowledge elements extracted through the Lexicon-LSTM model.Design/methodology/approachEight research themes were selected for experiment, with a large-scale (N = 11,625) dataset of research papers from the China National Knowledge Infrastructure (CNKI) database constructed. And it is complemented with multiple corpora. Knowledge elements were extracted through a Lexicon-LSTM model. A subject knowledge graph is constructed to support the searching and classification of knowledge elements. An interdisciplinary-weighted average citation index space was constructed for measuring the interdisciplinary characteristics and contributions based on knowledge elements.FindingsThe empirical research shows that the Lexicon-LSTM model has superiority in the accuracy of extracting knowledge elements. In the field of LIS, the interdisciplinary diversity indicator showed an upward trend from 2011 to 2021, while the disciplinary balance and difference indicators showed a downward trend. The knowledge elements of theory and methodology could be used to detect and measure the interdisciplinary characteristics and contributions.Originality/valueThe extraction of knowledge elements facilitates the discovery of semantic information embedded in academic papers. The knowledge elements were proved feasible for measuring the interdisciplinary characteristics and exploring the changes in the time sequence, which helps for overview the state of the arts and future development trend of the interdisciplinary of research theme in LIS.

A multi-scale graph embedding method via multiple corpora

Graph embedding aims at learning continuous vector representations for graphs which is crucial for graph analytics. Natural Language Process (NLP)-based graph embedding methods build corpus for graph data by treating substructures as words and then use NLP models to learn graph embeddings. However, the size difference and data redundancy among substructures are less explored in the built corpora. To mitigate this problem, we propose an unsupervised multi-scale graph embedding method. To be a specific, we first build multiple graph corpora for a graph dataset, where each corpus only contains substructures of specific granularity. Then, we extend a document embedding model to each graph corpus to obtain graph embeddings of different scales. At last, we obtain the final multi-scale embedding of a graph by pooling its multiple embeddings. Comprehensive experiments on real graph datasets indicate that the proposed method obtains competitive results with state-of-the-arts, and is superior to some classic graph kernels and graph embedding methods on six out of ten benchmark datasets.

JCBIE: a joint continual learning neural network for biomedical information extraction

Extracting knowledge from heterogeneous data sources is fundamental for the construction of structured biomedical knowledge graphs (BKGs), where entities and relations are represented as nodes and edges in the graphs, respectively. Previous biomedical knowledge extraction methods simply considered limited entity types and relations by using a task-specific training set, which is insufficient for large-scale BKGs development and downstream task applications in different scenarios. To alleviate this issue, we propose a joint continual learning biomedical information extraction (JCBIE) network to extract entities and relations from different biomedical information datasets. By empirically studying different joint learning and continual learning strategies, the proposed JCBIE can learn and expand different types of entities and relations from different datasets. JCBIE uses two separated encoders in joint-feature extraction, hence can effectively avoid the feature confusion problem comparing with using one hard-parameter sharing encoder. Specifically, it allows us to adopt entity augmented inputs to establish the interaction between named entity recognition and relation extraction. Finally, a novel evaluation mechanism is proposed for measuring cross-corpus generalization errors, which was ignored by traditional evaluation methods. Our empirical studies show that JCBIE achieves promising performance when continual learning strategy is adopted with multiple corpora.

Analysis and Application of the Business English Translation Query and Decision Model with Big Data Corpus

This paper aims to build an English translation query and decision support model using big data corpus and applies it to business English translation. Firstly, the existing convolutional network is improved by using depth-separable convolution, and the input statements are mapped to the depth feature space. Secondly, the attentional mechanism is used to enhance the expressive ability of input sentences in deep feature space. Then, considering the sequential relationship, use long short-term memory (LSTM) neural network as a decoder block to generate the corresponding translation of the input sentence. Finally, nonparametric metric learning module is used to improve the model in an end-to-end way. Wide range of experiments on the multiple corpora have shown the proposed model has better real-time performance while maintaining high precision in translation and query, and it has a certain practical application value.

A comparative study of pre-trained language models for named entity recognition in clinical trial eligibility criteria from multiple corpora

BackgroundClinical trial protocols are the foundation for advancing medical sciences, however, the extraction of accurate and meaningful information from the original clinical trials is very challenging due to the complex and unstructured texts of such documents. Named entity recognition (NER) is a fundamental and necessary step to process and standardize the unstructured text in clinical trials using Natural Language Processing (NLP) techniques.MethodsIn this study we fine-tuned pre-trained language models to support the NER task on clinical trial eligibility criteria. We systematically investigated four pre-trained contextual embedding models for the biomedical domain (i.e., BioBERT, BlueBERT, PubMedBERT, and SciBERT) and two models for the open domains (BERT and SpanBERT), for NER tasks using three existing clinical trial eligibility criteria corpora. In addition, we also investigated the feasibility of data augmentation approaches and evaluated their performance.ResultsOur evaluation results using tenfold cross-validation show that domain-specific transformer models achieved better performance than the general transformer models, with the best performance obtained by the PubMedBERT model (F1-scores of 0.715, 0.836, and 0.622 for the three corpora respectively). The data augmentation results show that it is feasible to leverage additional corpora to improve NER performance.ConclusionsFindings from this study not only demonstrate the importance of contextual embeddings trained from domain-specific corpora, but also shed lights on the benefits of leveraging multiple data sources for the challenging NER task in clinical trial eligibility criteria text.

Summarization assessment methodology for multiple corpora using queries and classification for functional evaluation

Extractive summarization is an important natural language processing approach used for document compression, improved reading comprehension, key phrase extraction, indexing, query set generation, and other analytics approaches. Extractive summarization has specific advantages over abstractive summarization in that it preserves style, specific text elements, and compound phrases that might be more directly associated with the text. In this article, the relative effectiveness of extractive summarization is considered on two widely different corpora: (1) a set of works of fiction (100 total, mainly novels) available from Project Gutenberg, and (2) a large set of news articles (3000) for which a ground truthed summarization (gold standard) is provided by the authors of the news articles. Both sets were evaluated using 5 different Python Sumy algorithms and compared to randomly-generated summarizations quantitatively. Two functional approaches to assessing the efficacy of summarization using a query set on both the original documents and their summaries, and using document classification on a 12-class set to compare among different summarization approaches, are introduced. The results, unsurprisingly, show considerable differences consistent with the different nature of these two data sets. The LSA and Luhn summarization approaches were most effective on the database of fiction, while all five summarization approaches were similarly effective on the database of articles. Overall, the Luhn approach was deemed the most generally relevant among those tested.

A Diachronic Study on the Fine Categorization of Chinese Synonymous Category Classifier Constructions

Based on multiple corpora, this study combines qualitative research and quantitative research and takes the category classifier of ancient Chinese as the research object, with the aim to investigate the cognitive mechanism of fine categorization thinking and embodiment reflected by the Chinese nationality in the process of its construction. The results show that: 1) Metonymy plays a leading role in the formation of "X Zhong/Lei N" category classfier construction and is responsible for "concrete relevance" while metaphor plays a leading role in the evolution of its construction and is responsible for "abstract analogy". 2) The mechanism of constructions extension is not limited to the four types proposed by Goldberg: metaphor, polysemy, exemplification and subcategorization, but also involves metonymy in the extension of Chinese synonymous category classifier constructions.

Factors in Emotion Recognition With Deep Learning Models Using Speech and Text on Multiple Corpora

Emotion recognition performance of deep learning models is influenced by multiple factors such as acoustic condition, textual content, style of emotion expression (e.g. acted, natural), etc. In this paper, multiple factors are analysed by training and evaluating state-of-the-art deep learning models using the input modalities speech, text, and their combination across 6 emotional speech corpora. A novel deep learning model architecture is presented that further improves the state-of-the-art in multimodal emotion recognition with speech and text on the IEMOCAP corpus. Results from models trained on individual corpora show that combining speech and text improves performance only on corpora where the text of utterances varies across different emotions, while it reduced performance on corpora with fixed text expressed in different emotions, where the speech-only models performed better. Further, cross-corpus investigations are presented to understand the robustness to changing acoustic and textual content. Results show that models perform significantly better in matched conditions in particular single corpus models perform better than multi-corpus models, with the latter showing a tendency to be more robust to acoustic variations, while performance still depends on characteristics of both training corpora and test corpus.

Generalizing Cross-Document Event Coreference Resolution Across Multiple Corpora

Cross-document event coreference resolution (CDCR) is an NLP task in which mentions of events need to be identified and clustered throughout a collection of documents. CDCR aims to benefit downstream multidocument applications, but despite recent progress on corpora and system development, downstream improvements from applying CDCR have not been shown yet. We make the observation that every CDCR system to date was developed, trained, and tested only on a single respective corpus. This raises strong concerns on their generalizability—a must-have for downstream applications where the magnitude of domains or event mentions is likely to exceed those found in a curated corpus. To investigate this assumption, we define a uniform evaluation setup involving three CDCR corpora: ECB+, the Gun Violence Corpus, and the Football Coreference Corpus (which we reannotate on token level to make our analysis possible). We compare a corpus-independent, feature-based system against a recent neural system developed for ECB+. Although being inferior in absolute numbers, the feature-based system shows more consistent performance across all corpora whereas the neural system is hit-or-miss. Via model introspection, we find that the importance of event actions, event time, and so forth, for resolving coreference in practice varies greatly between the corpora. Additional analysis shows that several systems overfit on the structure of the ECB+ corpus. We conclude with recommendations on how to achieve generally applicable CDCR systems in the future—the most important being that evaluation on multiple CDCR corpora is strongly necessary. To facilitate future research, we release our dataset, annotation guidelines, and system implementation to the public.1

Contextualizing /s/ retraction: Sibilant variation and change in Washington D.C. African American Language

AbstractRecent work has demonstrated an ongoing change across varieties of English in which /s/ retracts before consonants, particularly before /tɹ/ clusters (e.g., Lawrence, 2000; Shapiro, 1995; Stuart-Smith et al., 2019). Much of this work has focused on the social and linguistic distributions of /stɹ/ within single communities, without an examination of the broader sibilant space (e.g., /s/ and /ʃ/). Meanwhile, analyses across multiple corpora have shown that /s/ and /ʃ/ also show within-community variability, beyond /stɹ/ contexts (Stuart-Smith et al., 2019, 2020). Intersecting these approaches, this paper explores sibilant variation and change across /stɹ/, /s/, and /ʃ/ using a corpus of Washington D.C. African American Language (AAL). Results indicate that /stɹ/-retraction is a stable variant in this variety of AAL and /s/ and /ʃ/ show evidence of socially stratified variation and change. Overall, this paper demonstrates the need to examine the sibilant space more holistically when examining changes in /stɹ/.

Automated accent rating using deep neural networks

Automated accentedness rating has the potential to improve many human-computer interactions involving speech, including the adaptation of automatic speech recognition or other artificial intelligence models to the speaker's accent. Accent ratings may also be used as a metric by which language learners can quantify their progress. This study employs bidirectional long short-term memory layers in a neural network to predict human ratings of the accentedness of recorded speech. Speech data are extracted in 5-s segments from over 2000 first- and second-language English speakers from multiple corpora. Human ratings are obtained in an online experiment where participants rate the accentedness of a given speech recording on a 9-point Likert scale. Mel-frequency cepstral coefficients and mel-filterbank energy features are tested as speech input representations for the neural network. When inference is tested using 10-fold cross validation, the mean correlation between the model’s predictions and human ratings is high (r = 0.74). While previous methods attained a similar correlation by automatically comparing speech that has been transcribed [Wieling et al., Lang. Dyn. Chang. 4, 253–269 (2014)] or by making accent-specific Gaussian mixture models [Cheng et al., Interspeech 2013 (2013), pp. 2574–2578], the present model requires no transcription and can perform accent-general inference.

GnRH or estradiol benzoate combination with CIDR improves in-vivo embryo production in bovines (Bos indicus and Bos taurus) under subtropics.

Multiple Ovulation and Embryo Transfer (MOET) technology is a potential technique to upgrade livestock species’ genetics. The varied response to super-stimulatory treatments remains one of the limiting factors to this technology’s widespread use. The present study was aimed to improve the superovulation response and in-vivo embryo production by using controlled internal drug release (CIDR)-GnRH or CIDR-EB (Estradiol Benzoate) along with conventional superovulation protocol in Holstein Frisian (HF): Bos taurus; n = 42) and Crossbred (XB: Cholistani (Bos indicus) × HF; n = 28) cows. In the CIDR-GnRH/CIDR-EB treatment, CIDR was implanted in the cows after confirming the presence of a corpus luteum (CL) on the 8th day after estrus. 2 ml GnRH (Lecirelin acetate 0.0262 mg/ml) or 2 mg EB was also administered in CIDR-GnRH/CIDR-EB groups, respectively. Both groups were given super-stimulatory treatment from the 11th day after estrus (FSH in tapering doses twice a day for four consecutive days). On day 13, two doses of 2 ml prostaglandin (75 µg/ml of dextrorotatory cloprostenol) were administered (am: pm), and CIDR was removed the following day. Two artificial inseminations (AI) of the cows were performed (12 h apart) on the 15th day. No CIDR and GnRH/E.B were given in the control group, but the remaining superovulation protocol was the same. Later on, seven days after the first AI, non-surgical embryo flushing was done. The transferable embryos produced from three different superovulation protocols were then transferred into the recipient cows (n = 90) for determining their fertility. Statistical analysis revealed that the number of super-estrus follicles (SEF), multiple corpora lutea (MCL), ovulation/fertilization percentage, fertilized structures recovered (FSR), and transferable embryos (TEs) remained significantly higher (p < 0.05), and days taken for return to estrus (RTE) after embryo collection remained significantly lower (p < 0.05) in CIDR-GnRH (n = 18) and CIDR-EB (n = 15) groups as compared to the control (n = 37). The comparison between XB and HF cows revealed that the TEs production in CIDR-GnRH (XB = 5 vs HF = 13) and CIDR-EB (XB = 6 vs HF = 9) based superovulation protocols were 11.60 ± 4.08 vs 04.31 ± 0.98 and 09.33 ± 1.78 vs 05.22 ± 1.36, respectively. TEs production in XB cows (n = 5) of the CIDR-GnRH group was significantly higher (11.60 ± 4.08) than other groups. On the other hand, the days taken for RTE after embryo collection remained significantly lower (p < 0.05) in HF cows of treatment groups. However, the fertility of TEs was neither affected significantly (p > 0.05) by the superovulation protocol used nor by breed differences among donor cows. In conclusion, using CIDR-GnRH or CIDR-EB along with conventional superovulation protocol may enhance the efficiency of MOET programs in cattle. Furthermore, XB donor cows demonstrated a better performance than HF donor cows under subtropical conditions.

P–751 Immediate versus postponed frozen-thawed embryo transfer after IVF/ICSI: a systematic review and meta-analysis

Abstract Study question Can frozen embryo transfer (FET) be offered immediately after a stimulated IVF/ICSI cycle without compromising live birth rate (LBR)? Summary answer FET in the menstrual cycle immediately following the stimulated IVF/ICSI cycle was associated with a slightly higher LBR compared to standard postponed FET. What is known already It is standard clinical practice to postpone FET for at least one menstrual cycle following a failed fresh transfer or a freeze-all cycle. This practice is thought to minimize any possible residual negative effect of ovarian stimulation, with excessive steroid levels and multiple corpora lutea, on the resumption of a normal ovulatory cycle and receptivity of the endometrium. Even so, elective deferral of FET is an empirical strategy based on suggestions rather than solid scientific evidence and may unnecessarily delay time to pregnancy, causing frustration and decreased quality of life to couples. Study design, size, duration Systematic review and meta-analysis according to PRISMA guidelines. Original studies on subfertile women aged 18–46 with any indication for treatment with IVF/ICSI investigating the timing of FET after IVF/ICSI were included. Intervention was defined as FET in the menstrual cycle immediately following the stimulated IVF/ICSI cycle. Comparator was defined as FET in the second or subsequent menstrual cycle following IVF/ICSI. Risk of bias was assessed using Robins-I and quality of evidence using GRADE. Participants/materials, setting, methods PubMed (MEDLINE) and EMBASE databases were searched for MeSH and Emtree terms, as well as text words related to timing of FET, up to March 2020. There were no limitations regarding year of publication or duration of follow-up but to English language. The primary outcome was LBR. Secondary outcomes were implantation rate, pregnancy rate, clinical pregnancy rate (CPR), time-to-pregnancy, miscarriage rate (MR), cycle cancellation rate and patient wellbeing. Main results and the role of chance Out of 4124 search results, 15 studies were included in the review. Studies reporting adjusted odds ratios (aOR) for LBR, CPR and MR were included in meta-analyses. All studies (n = 15) were retrospective cohort studies involving a total of 6,304 immediate FET cycles and 13,851 postponed FET cycles including 8,019 matched controls. Twelve studies of very low to moderate quality reported no difference in LBR with immediate versus postponed FET. Two studies of moderate quality reported a statistically significant increase in LBR with immediate FET and one small study of very low quality reported better LBR with postponed FET. Trends in rates of secondary outcomes followed trends in LBR regarding timing of FET. The meta-analyses showed a significant advantage of immediate FET (n = 2,076) compared to postponed FET (n = 3,833), with a pooled aOR of 1.20 (95% CI 1.01–1.44) for LBR and a pooled aOR of 1.22 (95% CI 1.07–1.39) for CPR. Limitations, reasons for caution: Limitations include the retrospective design and heterogeneity of studies included, limiting comparison and pooling of data. With little transparency regarding cancellation rates, the risk of selection bias is apparent. Further, confounding by embryo quality is a limitation. Small sample sizes are a limitation to subgroup meta-analyses. Wider implications of the findings: The standard clinical practice of postponing FET for at least one menstrual cycle following a failed fresh transfer or a freeze-all cycle may not be best clinical practice. Randomized controlled trials including data on cancellation rates are highly needed to provide high grade evidence regarding clinical practice and patient counseling. Trial registration number Not applicable

P-751 Immediate versus postponed frozen-thawed embryo transfer after IVF/ICSI: a systematic review and meta-analysis

Abstract Study question Can frozen embryo transfer (FET) be offered immediately after a stimulated IVF/ICSI cycle without compromising live birth rate (LBR)? Summary answer FET in the menstrual cycle immediately following the stimulated IVF/ICSI cycle was associated with a slightly higher LBR compared to standard postponed FET. What is known already It is standard clinical practice to postpone FET for at least one menstrual cycle following a failed fresh transfer or a freeze-all cycle. This practice is thought to minimize any possible residual negative effect of ovarian stimulation, with excessive steroid levels and multiple corpora lutea, on the resumption of a normal ovulatory cycle and receptivity of the endometrium. Even so, elective deferral of FET is an empirical strategy based on suggestions rather than solid scientific evidence and may unnecessarily delay time to pregnancy, causing frustration and decreased quality of life to couples. Study design, size, duration Systematic review and meta-analysis according to PRISMA guidelines. Original studies on subfertile women aged 18-46 with any indication for treatment with IVF/ICSI investigating the timing of FET after IVF/ICSI were included. Intervention was defined as FET in the menstrual cycle immediately following the stimulated IVF/ICSI cycle. Comparator was defined as FET in the second or subsequent menstrual cycle following IVF/ICSI. Risk of bias was assessed using Robins-I and quality of evidence using GRADE. Participants/materials, setting, methods PubMed (MEDLINE) and EMBASE databases were searched for MeSH and Emtree terms, as well as text words related to timing of FET, up to March 2020. There were no limitations regarding year of publication or duration of follow-up but to English language. The primary outcome was LBR. Secondary outcomes were implantation rate, pregnancy rate, clinical pregnancy rate (CPR), time-to-pregnancy, miscarriage rate (MR), cycle cancellation rate and patient wellbeing. Main results and the role of chance Out of 4124 search results, 15 studies were included in the review. Studies reporting adjusted odds ratios (aOR) for LBR, CPR and MR were included in meta-analyses. All studies (n = 15) were retrospective cohort studies involving a total of 6,304 immediate FET cycles and 13,851 postponed FET cycles including 8,019 matched controls. Twelve studies of very low to moderate quality reported no difference in LBR with immediate versus postponed FET. Two studies of moderate quality reported a statistically significant increase in LBR with immediate FET and one small study of very low quality reported better LBR with postponed FET. Trends in rates of secondary outcomes followed trends in LBR regarding timing of FET. The meta-analyses showed a significant advantage of immediate FET (n = 2,076) compared to postponed FET (n = 3,833), with a pooled aOR of 1.20 (95% CI 1.01-1.44) for LBR and a pooled aOR of 1.22 (95% CI 1.07-1.39) for CPR. Limitations, reasons for caution Limitations include the retrospective design and heterogeneity of studies included, limiting comparison and pooling of data. With little transparency regarding cancellation rates, the risk of selection bias is apparent. Further, confounding by embryo quality is a limitation. Small sample sizes are a limitation to subgroup meta-analyses. Wider implications of the findings The standard clinical practice of postponing FET for at least one menstrual cycle following a failed fresh transfer or a freeze-all cycle may not be best clinical practice. Randomized controlled trials including data on cancellation rates are highly needed to provide high grade evidence regarding clinical practice and patient counseling. Trial registration number not applicable

Application of active learning algorithm in handwriting recognition numbers

Active learning is very suitable for many problems in natural language processing, where unlabeled data may be abundant, but annotation is slow and expensive. This article aims to illustrate some active learning methods for handwritten digit recognition tasks, such as the least confidence and entropy methods. We investigated the previously used sequence model query selection strategies and used some selection strategies for sample labeling in handwritten digit recognition. We also conduct a large-scale empirical comparison of using multiple corpora, which shows that our proposed method improves the technical level.

Global estimates of syntactic alignment in adult and child utterances during interaction: NLP estimates based on multiple corpora

The study of how conversational partners align language behavior has been given much attention in psychology, linguistics and sociology. Prominent frameworks for investigating this type of cooperative behavior include dialogic resonance, the interactive alignment model, abstract structural priming, and format tying. Yet, we do not have estimates of the overall amount of discourse alignment in human conversation. Obtaining estimates is challenging because alignment is complex, occurring at many different levels, and the resources needed to manually annotate large corpora to measure all aspects of discourse alignment restrict rapid progress on the issue. However, recent NLP research has developed systems to estimate alignment at least at the lexical and syntactic levels of conversation. While such systems come with error rates and lack the precision of manual annotation, interesting findings are emerging such as a recent study reporting grammatical alignment levels in the British National Corpus (BNC) was not particularly high, possibly what would be expected by chance. This is a controversial paper. The current study therefore, replicates the study in the recently released BNC 2014 spoken, finding in this corpus syntactic similarity is also close to the alignment one would get by randomly pairing conversational turns. The study goes further, finding the most frequent similarity level between utterances is zero, and that there exist no convincing cases of abstract structural alignment. Extending the research to child language corpora, grammatical alignment between children and adults is slightly above chance, though the most frequent level of syntactic similarity is zero, with no convincing cases of an abstract structure used by a child because their adult interlocutor had primed them. The findings might mediate theory-building that assumes discourse alignment at the grammatical level is widespread throughout conversation between adults or adults and children.

Cross-Corpus Speech Emotion Recognition Based on Few-Shot Learning and Domain Adaptation

Within a single speech emotion corpus, deep neural networks have shown decent performance in speech emotion recognition. However, the performance of the emotion recognition based on data-driven learning methods degrades significantly for the cross-corpus scenario. To relieve this issue without any labeled samples from the target domain, we propose a cross-corpus speech emotion recognition based on few-shot learning and unsupervised domain adaptation, which is trained to learn the class (emotion) similarity from the source domain samples adapted to the target domain. In addition, we utilize multiple corpora in training to enhance the robustness of the emotion recognition to the unseen samples. Experiments on emotional speech corpora with three different languages showed that the proposed method outperformed other approaches.

Progesterone-responsive vaginal leiomyoma and hyperprogesteronemia due to ovarian luteoma in an older bitch

BackgroundThis is the first report about a vaginal leiomyoma concomitant with an ovarian luteoma in a bitch.Case presentationA 11-year-old intact female Labrador retriever was referred because of anuria, constipation and protrusion of a vaginal mass through the vulvar commissure. The bitch had high serum progesterone concentration (4.94 ng/ml). Because of the possibility of progesterone responsiveness causing further increase of the vaginal mass and since the bitch was a poor surgical candidate a 10 mg/kg aglepristone treatment was started SC on referral day 1. A computerized tomography showed a 12.7 × 6.5 × 8.3 cm mass causing urethral and rectal compression, ureteral dilation and hydronephrosis. A vaginal leiomyoma was diagnosed on histology. As serum progesterone concentration kept increasing despite aglepristone treatment, a 0.02 ng/mL twice daily IM alfaprostol treatment was started on day 18. As neither treatment showed remission of clinical signs or luteolysis, ovariohysterectomy was performed on referral day 35. Multiple corpora lutea were found on both ovaries. On histology a luteoma was diagnosed on the left ovary. P4 levels were undetectable 7 days after surgery. Recovery was uneventful and 12 weeks after surgery tomography showed a reduction of 86.7% of the vaginal mass. The bitch has been in good health and able to urinate without any complication ever since.ConclusionsThis case demonstrates the importance of identifying progesterone related conditions as well as the importance of judiciously using a combined medical and surgical approach.