Search

Abstract In recent years, there has been a significant increase in the design of neural network models for solving math word problems (MWPs). These neural solvers have been designed with various architectures and evaluated on diverse datasets, posing challenges in fair and effective performance evaluation. This paper presents a comparative study of representative neural solvers, aiming to elucidate their technical features and performance variations in solving different types of MWPs. Firstly, an in-depth technical analysis is conducted from the initial deep neural solver DNS to the state-of-the-art GPT-4. To enhance the technical analysis, a unified framework is introduced, which comprises highly reusable modules decoupled from existing MWP solvers. Subsequently, a testbed is established to conveniently reproduce existing solvers and develop new solvers by combing these reusable modules, and finely regrouped datasets are provided to facilitate the comparative evaluation of the designed solvers. Then, comprehensive testing is conducted and detailed results for eight representative MWP solvers on five finely regrouped datasets are reported. The comparative analysis yields several key findings: (1) Pre-trained language model-based solvers demonstrate significant accuracy advantages across nearly all datasets, although they suffer from limitations in math equation calculation. (2) Models integrated with tree decoders exhibit strong performance in generating complex math equations. (3) Identifying and appropriately representing implicit knowledge hidden in problem texts is crucial for improving the accuracy of math equation generation. Finally, the paper also discusses the major technical challenges and potential research directions in this field. The insights gained from this analysis offer valuable guidance for future research, model development, and performance optimization in the field of math word problem solving.

Developing Arithmetic Word Problem (AWP) -solving algorithms has recently become one of the hottest research areas because it can simultaneously advance general artificial intelligence and the application of AI technology in education. This paper presents a novel algorithm for solving AWPs by synergizing Large Language Models (LLMs) with the Scene-Aware Syntax–Semantics (S2) method. The innovation of this algorithm lies in leveraging the LLM to divide problems into multiple scenes, thereby enhancing the relation-flow approach in the processes of relation extraction and reasoning. Our algorithm consists of three components: scene decomposer, relation extractor, and symbolic solver. In the scene decomposer, we propose the Chain-Of-Scene (COS) method. It dynamically constructs prompts for the LLM using a retrieval-augmented strategy, thus enabling the chain-formed generation of scenes from the input problem. In the relation extractor, we introduce the Scene-Aware S2 method, which utilizes syntax–semantics models to match the text within each scene and convert it into relations. This allows for the efficient and accurate extraction of explicit and implicit relations. Finally, a symbolic solver is employed to reason through the set of relations to derive the solution. Experimental results on six authoritative datasets demonstrate that the proposed algorithm achieves an average solving accuracy of 90.4%, outperforming the State-Of-The-Art (SOTA) algorithm by 1.1%. The case study further illustrates that it outputs more reliable solutions than baseline algorithms. These findings have significant implications for promoting smart education and developing personalized intelligent tutoring systems.

A tutorable algorithm for solving text-diagram function (TDF) problems is an essential technology for building the humanoid tutorial service since ”function” is a core portion of mathematics. Solving TDF problems encounters challenges in understanding diagrams, representing the functions of being compound objects, and solving the mixture group of functions and universal relations. To address these challenges, this paper proposes a relation-centric algorithm, leveraging on making breakthrough in handling function, understanding diagram, and generating tutorable solution. The proposed algorithm comprises two phases: problem understanding and symbolic solver. In the problem understanding, it proposes a S2P (Syntax Semantics with Period) model method of acquiring relations from text and a L2 (Line Segment with Labels Pattern) model method of acquiring relations from diagrams. To get the problem fully understood, this phase acquires both universal relations and period relations. In the symbolic solver, the function is first built from the acquired relations. Then an equation-function interaction method is created to solve a mixture system of relations and functions. The developed algorithm is the first one for solving text-diagram function problems. Experimental results show that the proposed algorithm not only has high accuracies of 74.5% on Math23KtoF and 80.8% on TnD1K datasets, but also can produce the tutorable solutions of TDF problems.

This paper presents a vector relation-centric algorithm for solving arithmetic word problems (AWPs), which uses vector relation acquisition and scene knowledge to ensure the performances of problem understanding and symbolic solver correspondingly. The vector relation acquisition procedure builds on the synergy of the vector syntax-semantics method and the deep neural miner. Compared with the syntax-semantics method, the vector syntax-semantics method decreases not only the number of models but also semantic ambiguities and computational costs. For the scene knowledge, this paper proposes a scene-aware symbolic solver which infers relations obeying scene rules to decrease the occurrences of unwanted operations. Experimental results show that the proposed algorithm is superior to the high-performance baseline algorithm in both accuracy and computational cost. In accuracy, the proposed algorithm increased the accuracy by 3.9% on the sum of the three scene subsets due to the use of the scene knowledge and vector computing; as a result, it increased the accuracy by 0.5% on the sum of six authoritative datasets. In computational cost, the proposed algorithm decreased the computing cost by more than 50%. Thus, this paper makes a significant contribution to developing instruments for solving AWPs.

Abstract This paper presents a relation-centric algorithm for solving arithmetic word problems (AWPs) by synergizing a syntax-semantics extractor for extracting explicit relations, and a neural network miner for mining implicit relations. This is the first algorithm that has a specific component to acquire implicit knowledge items for solving AWPs. This paper proposes a three-phase scheme to decompose the challenging task of designing an algorithm for solving AWPs into three smaller tasks. The first phase proposes a state-action paradigm; the second phase instantiates the paradigm into a relation-centric approach; and the third phase implements a relation-centric algorithm for solving AWPs. There are two main steps in the proposed algorithm: problem understanding and symbolic solver. By adopting the relation-centric approach, problem understanding becomes a task of relation acquisition. For conducting the task of relation acquisition, a relaxed syntax-semantics method first extracts a group of explicit relation candidates. In parallel, a neural network miner acquires implicit relation candidates. The miner computes the vectors encoded by BERT to determine which implicit relations should be added. Thus, problem understanding can acquire both explicit relations and implicit relations, which addresses the challenge of building a problem understanding method that can acquire all the knowledge items to find the solution. In the subsequent step of symbolic solver, a fusion procedure forms a distilled set of relations from all the candidates by discarding unnecessary relations. Experimentation on nine benchmark datasets validates the superiority of the proposed algorithm that outperforms the state-of-the-art algorithms.

We propose that despite the general positive impact of maintaining political connections, internet utilization generates a disconnecting effect on the impacts of political connections on firm innovation. Using a unique survey dataset of Chinese private enterprises, the present study found that private firm owners’ internet use moderates the relationship between political connections and firm innovation. The present study therefore enriches the literature on the effects of political connections and resource-based view by discovering that internet technology disconnects firms’ political connections and substitutes for the function of political connections on information search and resource acquisition.

Sex differences in academic behaviours have long been a concern for adolescent developmental researches. Evidences concerning sex differences in academic seeking help (AHS) behaviours are inconsistent. This paper aims to use meta-analysis to determine whether there is sex difference in AHS behaviours. After extensive retrieval and rigorous inclusion criteria, a total of 47 qualifying studies in the last two decades are identified (covering 27,149 students from grade school through to college). Several findings are standing out. Firstly, the effect size of sex differences in AHS behaviours is - 0.107 (p=0.048), indicating that school girls are more active and more adaptive in AHS behaviours than school boys. When further dividing AHS behaviours into different forms, the statistically significant sex differences are mainly embodied in instrumental HS, and no statistically significant sex differences are observed in executive HS. Secondly, the sex gap is narrowing with the increase of grade levels, but an abnormal rebound is observed in high school stage. Thirdly, there are statistically significant sex differences in instrumental HS for Asian students and executive HS for North American students. Last but not least, the sex gap reported in papers published in Chinese is smaller than that in English. These findings help to reveal the basic mechanism of sex differences in AHS behaviours and provide a basis for the development of effective AHS strategies.

Considered as a key component of human capital, mental health has drawn substantial scholarly attention for its effect on people's health status and economic outcome. When facing the challenge of stress, people's heterogeneity in cognitive ability and non-cognitive ability causes difference in patterns of coping, resulting in different manifestations in mental health. Previous researches have shown that cognitive and non-cognitive abilities have positively direct or indirect effects on mental health, but few studies research their role of coping with air pollution. We used the China Family Panel Survey (CFPS) and matched individual data with county or district level PM2.5 information from NASA. The study found that air pollution has negative effect on mental health with every increase of 1μg/m3 in PM2.5 deteriorating mental health by 0.038 standard deviation, which is the total effect of air pollution. However, the direct effect of air pollution on mental health will decrease to 0.028 in absolute value when considering mediating effects. By employing different approaches, we found positive mediating effects via cognitive ability and non-cognitive ability. Individuals with high cognitive and non-cognitive abilities are able to accurately diagnose problems and select the optimal coping strategies, thus restoring positive mental health.

In the field of pedestrian dead reckoning (PDR), the zero velocity update (ZUPT) method with an inertial measurement unit (IMU) is a mature technology to calibrate dead reckoning. However, due to the complex walking modes of different individuals, it is essential and challenging to determine the ZUPT conditions, which has a direct and significant influence on the tracking accuracy. In this research, we adopted an adaptive zero velocity update (AZUPT) method based on convolution neural networks to classify the ZUPT conditions. The AZUPT model was robust regardless of the different motion types of various individuals. AZUPT was then implemented on the Zynq-7000 SoC platform to work in real time to validate its computational efficiency and performance superiority. Extensive real-world experiments were conducted by 60 different individuals in three different scenarios. It was demonstrated that the proposed system could work equally well in different environments, making it portable for PDR to be widely performed in various real-world situations.