Chinese Comma Disambiguation in Math Word Problems Using SMOTE and Random Forests

Natural language understanding technologies play an essential role in automatically solving math word problems. In the process of machine understanding Chinese math word problems, comma disambiguation, which is associated with a class imbalance binary learning problem, is addressed as a valuable instrument to transform the problem statement of math word problems into structured representation. Aiming to resolve this problem, we employed the synthetic minority oversampling technique (SMOTE) and random forests to comma classification after their hyperparameters were jointly optimized. We propose a strict measure to evaluate the performance of deployed comma classification models on comma disambiguation in math word problems. To verify the effectiveness of random forest classifiers with SMOTE on comma disambiguation, we conducted two-stage experiments on two datasets with a collection of evaluation measures. Experimental results showed that random forest classifiers were significantly superior to baseline methods in Chinese comma disambiguation. The SMOTE algorithm with optimized hyperparameter settings based on the categorical distribution of different datasets is preferable, instead of with its default values. For practitioners, we suggest that hyperparameters of a classification models be optimized again after parameter settings of SMOTE have been changed.

Handedness Development: A Model for Investigating the Development of Hemispheric Specialization and Interhemispheric Coordination

The author presents his perspective on the character of science, development, and handedness and relates these to his investigations of the early development of handedness. After presenting some ideas on what hemispheric specialization of function might mean for neural processing and how handedness should be assessed, the neuroscience of control of the arms/hands and interhemispheric communication and coordination are examined for how developmental processes can affect these mechanisms. The author’s work on the development of early handedness is reviewed and placed within a context of cascading events in which different forms of handedness emerge from earlier forms but not in a deterministic manner. This approach supports a continuous rather than categorical distribution of handedness and accounts for the predominance of right-handedness while maintaining a minority of left-handedness. Finally, the relation of the development of handedness to the development of several language and cognitive skills is examined.

Handedness Development: A Model for Investigating the Development of Hemispheric Specialization and Interhemispheric Coordination

The author presents his perspective on the character of science, development, and handedness and relates these to his investigations of the early development of handedness. After presenting some ideas on what hemispheric specialization of function might mean for neural processing and how handedness should be assessed, the neuroscience of control of the arms/hands and interhemispheric communication and coordination are examined for how developmental processes can affect these mechanisms. The author’s work on the development of early handedness is reviewed and placed within a context of cascading events in which different forms of handedness emerge from earlier forms but not in a deterministic manner. This approach supports a continuous rather than categorical distribution of handedness and accounts for the predominance of right-handedness while maintaining a minority of left-handedness. Finally, the relation of the development of handedness to the development of several language and cognitive skills is examined.

Interactional Meta-Discourse Resources in Oliver Twist

For long, there has been debate over the appropriateness of using simplified literary texts in second language classrooms. In examining the simplified form, the main question is persuasion that is partly achieved through meta-discourse resources, which are “Self-reflective linguistic material referring to the evolving text and to the writer and the imagined reader of the text” (Hylan & Tse, 2004, p. 156). The present study compares the use of interactional meta-discourse resources (IMRs) in terms of the frequency and categorical distribution in the original copy of a novel (i.e., Oliver Twist) and its simplified counterparts. The corpus was analyzed based on the Hyland (2005) model. The frequency and categorical distribution of IMRs were calculated per 1,000 words, and the difference in their distribution was calculated using Chi-Square statistical analysis. The findings indicate a significant difference in the frequency of IMRs between the original and the simplified versions of the Dickensian novel, implying that despite having more complex syntax and semantics, the original novel seems to be more persuasive, at least on the part of IMRs, compared to its simplified counterparts. In terms of categorical distribution, there was no significant difference between them.

Topic Modeling in Embedding Spaces

Topic modeling analyzes documents to learn meaningful patterns of words. However, existing topic models fail to learn interpretable topics when working with large and heavy-tailed vocabularies. To this end, we develop the embedded topic model (etm), a generative model of documents that marries traditional topic models with word embeddings. More specifically, the etm models each word with a categorical distribution whose natural parameter is the inner product between the word’s embedding and an embedding of its assigned topic. To fit the etm, we develop an efficient amortized variational inference algorithm. The etm discovers interpretable topics even with large vocabularies that include rare words and stop words. It outperforms existing document models, such as latent Dirichlet allocation, in terms of both topic quality and predictive performance.

133 Test of minimum-intervention protocols for optimizing in vitro embryo production in bison

The study was done to determine whether minimal handling protocols for ovarian synchronization and ovarian superstimulation may be used to increase in vitro embryo production in bison. Ultrasound-guided cumulus-oocyte complex (COC) collection was done in a group of bison (n=23; random start) during the anovulatory season to synchronize new follicular wave emergence. The COC were classified morphologically (compact-good and -regular, expanded, denuded, degenerate) but not processed further. At the time of COC collection (Day 0), bison were assigned randomly to 3 groups and given 5mL of saline IM (non-superstimulated controls; n=11), 10 Armour units of pFSH (Antrin R10, Kyoritsu Seiyaku Corp., Tokyo, Japan) in 5mL of saline IM once per day from Day 0 to 2 (regular FSH; n=5), or 30 armour units of a sustained-release form of pFSH (Antrin R10Al, Kyoritsu Seiyaku Corp.) in 5mL of saline SC on Day 0 (long-acting FSH; n=7). On Day 4, a second COC collection was performed. Only compact COC were processed. The COC were matured in vitro for 25 to 28h at 38.8°C, fertilized (2×106 sperm mL−1) and co-incubated at 38.8°C in 5% CO2 for 18h. Presumptive zygotes were denuded and cultured at 38.8°C in 5% O2, 5% CO2 and 90% N2. Nominal data were compared by t-test and analysis of variance. Binomial data were compared among groups by chi-squared. There was no difference between the first (random) COC collection (n=23) and second collection (n=11 non-superstimulated controls) in the total number of follicles detected, but the distribution among size categories (3-4, 4-8, and >8mm) differed, i.e. fewer in the 3 to 4mm category at the time of the second COC collection (12.2±1.0v. 8.1±1.4; P<0.05). In the nonstimulated control group, there were no differences between the first and second COC collections in the number of follicles aspirated (12.7±1.0v. 10.4±1.5), number of COC collected (7.7±0.9v. 5.3±1.3), or in the categorical distribution of COC. At the second COC collection, the number of follicles in the >8mm category was greater in the regular FSH group than in the control or long-acting FSH groups (2.8±0.5v. 1.1±0.3, and 1.9±0.4, respectively; P<0.05), but no differences were detected in the number of follicles aspirated, COC collected, or in the categorical distribution of COC. The cleavage rate (of total oocytes submitted to in vitro maturation), recorded 2 days after IVF, was higher in the control group than in either the regular FSH or long-acting FSH groups [25/35 (71.4%), 7/28 (25.0%), 8/35 (22.8%); P<0.0001]. The freezable embryo production rate, recorded 7 days after IVF, was greater in the control group than in the regular FSH or long-acting FSH groups [19/35 (54.3%), 5/28 (17.9%), 5/35 (14.3%); P<0.01]. In conclusion, minimal-handling interventions used in the present study to increase embryo production in bison were not effective, likely as a result of the timing, frequency, and duration of superstimulation. A random start resulted in greater COC collection than collection 4 days after ovarian synchronization, and embryo production rates were greater in non-superstimulated bison. This work was supported by Parks Canada and Saskatchewan ADF. Antrin products donated by Kyoritsu Seiyaku Corp., Japan.

The Characterization of Biological Organization, Abstraction, and Novelty in Biomimetic Design

Through billions of years of evolution, a latent record of successful and failed design practices has developed in nature. The endeavors to exploit this record have resulted in numerous successful products in various fields of engineering, including, but not limited to, networking, propulsion, surface engineering, and robotics. In this work, a study of existing biomimetic designs has been carried out by categorizing the designs according to the biological organizational level, the abstraction level, and a novelty measure. The criterion of novelty has been used as a partial measure of the quality of bio-inspired and biomimetic designs already introduced, or ready to be introduced to the market. Through this review and categorization, we recognize patterns in existing biomimetic and bio-inspired products by analyzing their cross-categorical distribution. Using the distribution, we identify the categories which yield novel bio-inspired designs. We also examine the distribution to identify less explored areas of bio-inspired design. Additionally, this study is a step forward in aiding the practitioners of biomimetics in identifying the categories which yield the highest novelty products in their area of interest.

Association of anticardiolipin IgM antibody with preeclampsia

<p class="Abstract">The objective of this study was to evaluate association of anticardiolipin IgM antibody with preeclampsia among the 70 pregnant women having preeclampsia from March 2016 to February 2017. Equal number of age matched normotensive healthy pregnant women was selected as control. Demographic, obstetric and relevant laboratory data were collected and compared between the two groups. The level of anticardiolipin IgM antibody was significantly higher in preeclamptic women than in normal pregnant (mean 8.7 vs 5.8 U/mL, p&lt;0.001). Even in categorical distribution when the value was considered as positive by assessing a cut-off, cases had three times more chance to have positive level of anticardiolipin IgM antibody than controls (OR=3.3, 95% CI=1.1-9.6, p=0.046). Pearson’s correlation test revealed that both systolic and diastolic blood pressure had a positive correlation with the level of anticardiolipin IgM antibody. (p&lt;0.05). In conclusion, this study shows that anticardiolipin IgM antibody was associated with preeclampsia. The level of this anticardiolipin IgM antibody is directly proportional to the severity of preeclampsia and both systolic and diastolic blood pressure.</p>

The phonologisation of redundancy: length and quality in Welsh vowels

‘Phonologisation’ is a process whereby a phonetic phenomenon enters the phonological grammar and becomes conceptualised as the result of categorical manipulation of phonological symbols. I analyse the phonologisation of a predictable phonological pattern in Welsh, with particular attention to identifying criteria for whether phonologisation has occurred. I argue for a model where phonologisation experiences bottom-up and top-down biases. From the bottom up, there is pressure to phonologise phenomena with a categorical distribution; from the top down, there exist formal constraints on featural specification. I focus on the requirement for featural specifications to obey the Contrastivist Hypothesis, which denies that redundant features can be involved in phonological computation, in the context of a framework with emergent features. I suggest that the Contrastivist Hypothesis provides a useful boundary condition for emergent-feature theories, whilst independent phonologisation criteria provide contrastivist approaches with a more solid conceptual underpinning.