Chinese Conversation Structure

This paper aims to describe the features of Chinese conversation structure. Specifically speaking, the structure will be analyzed from the following four aspects: openings and pre-sequence, adjacency pairs, pre-closing and closing. Generally speaking, Chinese conversation structure is similar to English conversation structure. But still a lot of differences are found due to cultural factors.

Incorporating Linguistic Rules in Statistical Chinese Language Model for Pinyin-to-character Conversion

An N-gram Chinese language model incorporating linguistic rules is presented. By constructing elements lattice, rules information is incorporated in statistical frame. To facilitate the hybrid modeling, novel methods such as MI-based rule evaluating, weighted rule quantification and element-based n-gram probability approximation are presented. Dynamic Viterbi algorithm is adopted to search the best path in lattice. To strengthen the model, transformation-based error-driven rules learning is adopted. Applying proposed model to Chinese Pinyin-to-character conversion, high performance has been achieved in accuracy, flexibility and robustness simultaneously. Tests show correct rate achieves 94.81% instead of 90.53% using bi-gram Markov model alone. Many long-distance dependency and recursion in language can be processed effectively.

EVA2.0:Investigating Open-domain Chinese Dialogue Systems with Large-scale Pre-training

Large-scale pre-training has shown remarkable performance in building open-domain dialogue systems.However,previous works mainly focus on showing and evaluating the conversational performance of the released dialogue model,ignoring the discussion of some key factors towards a powerful human-like chatbot,especially in Chinese scenarios.In this paper,we conduct extensive experiments to investigate these under-explored factors,including data quality control,model architecture designs,training approaches,and decoding strategies.We propose EVA2.0,a large-scale pre-trained open-domain Chinese dialogue model with 2.8 billion parameters,and will make our models and codes publicly available.Automatic and human evaluations show that EVA2.0 significantly outperforms other open-source counterparts.We also discuss the limitations of this work by presenting some failure cases and pose some future research directions on large-scale Chinese open-domain dialogue systems.

Long-Term Clinical Effect of Tangyiping Granules(糖异平颗粒)on Patients with Impaired Glucose Tolerance

Objective： To evaluate the long-term clinical effect of Tangyiping Granules（糖异平颗粒, TYP） on patients with impaired glucose tolerance（IGT） to achieve normal glucose tolerance（NGT） and hence preventing them from conversion to diabetes mellitus（DM）. Methods： In total, 127 participants with IGT were randomly assigned to the control（63 cases, 3 lost to follow-up） and treatment groups（64 cases, 4 lost to follow-up） according to the random number table. The control group received lifestyle intervention alone, while the patients in the treatment group took orally 10 g of TYP twice daily in addition to lifestyle intervention for 12 weeks. The rates of patients achieving NGT or experiencing conversion to DM as main outcome measure were observed at 3, 12, and 24 months after TYP treatment. The secondary outcome measures included fasting plasma glucose（FPG）, 2-h postprandial plasma glucose（2h PG）, glycosylated hemoglobin（Hb A1c）, fasting insulin（FINS）, 2-h insulin（2hI NS）, homeostatic model assessment of insulin resistance（HOMA-IR）, blood lipid and patients＇ complains of Chinese medicine（CM） symptoms before and after treatment. Results： A higher proportion of the treatment group achieved NGT compared with the control group after 3-, 12- and 24-month follow-up（75.00% vs. 43.33%, 58.33% vs. 35.00%, 46.67% vs. 26.67%, respectively, P〈0.05）. The IGT to DM conversion rate of the treatment group was significantly lower than that of the control group at the end of 24-month follow-up（16.67% vs. 31.67%, P〈0.05）. Before treatment, FPG, 2h PG, Hb A1 c, FINS, 2h INS, HOMA-IR, triglyceride（TG）, total cholesterol, low- and high-density lipoprotein cholesterol levels had no statistical difference between the two groups（P〉0.05）. After treatment, the 2hP G, HbA 1c, HOMA-IR, and TG levels of the treatment group decreased significantly compared with those of the control group（P〈0.05）. CM symptoms such as exhaustion, irritability, chest tightness and breathless, spontaneous sweating, constipation, and dark thick and greasy tongue were significantly improved in the treatment group as compared with the control group（P〈0.05）. No severe adverse events occurred. Conclusion： TYP administered at the IGT stage with a disciplined lifestyle delayed IGT developing into type 2 DM.