Higher Retrieval Accuracy Research Articles

Medical Text Classification Based on an Optimized Machine Learning and External Semantic Resource

Automatic classification of texts is a well-known topic in natural language processing (NLP), and it involves categorizing unstructured texts into specific groups. Classification algorithms for a given problem are searched for, as well as the appropriate features to use as input. These models have been used for analysis of data with a large number of sparse dimensions. In medical field, categorization of medical texts is a difficult task because these types of texts contain several terminologies that define medical ideas and words. Medical data also lack proper sentence structure and do not adhere to the standard rules of natural language grammar. The main problem in this classification task is to choose an appropriate representation as input. To solve the representation problem, we propose a new extension model for medical text personification. Our model starts with the preprocessing of medical text. Then, we use external terminology resources to expand the text. After that, we use a combination of different text vector representations. To speed up the process, we try to increase the accuracy while selecting the optimal classification parameters. In this topic, we use an adaptive particle swarm optimization (PSO) algorithm to select the best parameters of a machine learning model. The PubMed, Hallmarks and AIM dataset were used to test our model. Our categorization text model outperforms the competition and provides much higher retrieval accuracy than previous models.

Retrieval of Soil Moisture Content Based on Multisatellite Dual-Frequency Combination Multipath Errors

Global navigation satellite system interferometric reflectometry (GNSS-IR) is a new type of microwave remote sensing technology that can measure soil moisture content (SMC). GNSS-IR soil moisture retrieval methods based on the satellite signal-to-noise ratio (SNR) and triple-frequency signal combination have the following shortcomings: SNR does not always exist in the original GNSS file, and the number of triple-frequency signal observation satellites is small, resulting in GNSS-IR soil moisture observation time resolution being low. Based on the above problems, in this study, we constructed a soil moisture inversion method based on multisatellite dual-frequency combined multipath error is proposed: the multipath error calculation model of dual-frequency carrier phase (L4 Ionosphere Free, L4_IF) and dual-frequency pseudorange (DFP) without ionospheric effect is constructed. We selected the data of the five epochs before and after the time point of the effective satellite period to construct the multipath error model and error equation, and we solved the delay phase for soil moisture retrieval. We verified the method using Plate Boundary Observatory (PBO) P041 site data. The results showed that the Pearson correlation coefficients (R) of L4_IF and DFP methods at P041 station are 0.97 and 0.91, respectively. To better verify the results’ reliability and the proposed method’s effectiveness, the soil moisture data of the MFLE station about 210 m away from P041 station are used as the verification data in this paper. The results showed that the delay phase solved by multipath error and soil moisture strongly correlate. Pearson correlation coefficients (R) of L4_IF and DFP methods at MFLE station are 0.93 and 0.86, respectively. In order to improve the inversion accuracy of GNSS-IR soil moisture, this paper constructs the prediction model of soil moisture by using the linear regression (ULR), back propagation neural network (BPNN) and radial basis function neural network (RBFNN), and evaluates the accuracy of each model. The results showed that the soil moisture retrieval method based on multisatellite dual-frequency combined multipath error can replace the traditional retrieval method and effectively improve the time resolution of GNSS-IR soil moisture estimation. To perform highly dynamic monitoring of soil moisture, higher retrieval accuracy can only be obtained with a small epoch multipath error.

Intelligent retrieval method of library document information based on hidden topic mining

In order to overcome the problems of retrieval accuracy and time-consuming of traditional document information retrieval methods, this paper designs an intelligent retrieval method of library document information based on hidden topic mining. Firstly, LDA model is used to mine the hidden topics of library document information, and then, based on the mining results, similarity degree of document information is calculated in inference network model. Finally, the Bayesian model is constructed in the sample space to retrieve the library literature information under the maximum retrieval space coverage. Experimental results show that, compared with traditional retrieval methods, the proposed method improves the retrieval accuracy significantly, with the highest retrieval accuracy reaching 99%, and the retrieval time is significantly reduced, indicating that the proposed method effectively improves the retrieval accuracy and timeliness.

Memristive Izhikevich Spiking Neuron Model and Its Application in Oscillatory Associative Memory.

The Izhikevich (IZH) spiking neuron model can display spiking and bursting behaviors of neurons. Based on the switching property and bio-plausibility of the memristor, the memristive Izhikevich (MIZH) spiking neuron model is built. Firstly, the MIZH spiking model is introduced and used to generate 23 spiking patterns. We compare the 23 spiking patterns produced by the IZH and MIZH spiking models. Secondly, the MIZH spiking model actively reproduces various neuronal behaviors, including the excitatory cortical neurons, the inhibitory cortical neurons, and other cortical neurons. Finally, the collective dynamic activities of the MIZH neuronal network are performed, and the MIZH oscillatory network is constructed. Experimental results illustrate that the constructed MIZH spiking neuron model performs high firing frequency and good frequency adaptation. The model can easily simulate various spiking and bursting patterns of distinct neurons in the brain. The MIZH neuronal network realizes the synchronous and asynchronous collective behaviors. The MIZH oscillatory network can memorize and retrieve the information patterns correctly and efficiently with high retrieval accuracy.

Case-based reasoning system for fault diagnosis of aero-engines

Fault diagnosis in aero-engines typically requires an experienced expert for understanding and detecting the cause of faults. However, accurate and quick identification of fault parts is difficult for maintenance crews owing to the complexity of aero-engines. In this study, we developed a case-based reasoning (CBR) system with a highly accurate novel similarity measure for fault diagnosis of aero-engines by retrieving similar fault cases. The proposed CBR system is established based on 143 cases with the knowledge of correctly diagnosed and successfully resolved aero-engine faults, which constitutes the first tentative case base in the field of aero-engine fault diagnosis. The proposed case similarity measure for fault diagnosis of aero-engines integrates three local similarity measures associated with different attributes, especially among which a tree-based semantic similarity measure is proposed to define the relationship between the fault part and fault mode based on a semantic graph incorporated into the aero-engine tree structure. The proposed CBR system is evaluated using the k-nearest neighbors algorithm and 5-fold cross-validation. The system exhibited high retrieval accuracy with all cases collected from real-world scenarios of aero-engine fault diagnosis. Our study shows great promise that the experience-based decisions yielded from the results can aid in aero-engine maintenance and support services.

An NDVI Retrieval Method Based on a Double-Attention Recurrent Neural Network for Cloudy Regions

NDVI is an important parameter for environmental assessment and precision agriculture that well-describes the status of vegetation. Nevertheless, the clouds in optical images often result in the absence of NDVI information at key growth stages. The integration of SAR and optical image features will likely address this issue. Although the mapping of different data sources is complex, the prosperity of deep learning technology provides an alternative approach. In this study, the double-attention RNN architecture based on the recurrent neural network (RNN) and attention mechanism is proposed to retrieve NDVI data of cloudy regions. Overall, the NDVI is retrieved by the proposed model from two aspects: the temporal domain and the pixel neighbor domain. The performance of the double-attention RNN is validated through different cloud coverage conditions, input ablation, and comparative experiments with various methods. The results conclude that a high retrieval accuracy is guaranteed by the proposed model, even under high cloud coverage conditions (R2 = 0.856, RMSE = 0.124). Using SAR images independently results in poor NDVI retrieval results (R2 = 0.728, RMSE = 0.141) with considerable artifacts, which need to be addressed with auxiliary data, such as IDM features. Temporal and pixel neighbor features play an important role in improving the accuracy of NDVI retrieval (R2 = 0.894, RMSE = 0.096). For the missing values of NDVI data caused by cloud coverage, the double-attention RNN proposed in this study provides a potential solution for information recovery.

Improved Accuracy of Animal Skin Pattern retrieval with CNN Model and Distance Metrics

This study describes the retrieval of animal skins with very diverse shapes and textures. With so many variations in each animal skin pattern, an appropriate and supported CNN model is needed as well as an appropriate distance matrix method to support retrieval performance. This research was conducted on 6 types of animal skin images. In experimenting to obtain this animal skin image, 4 CNN models were used, namely EfficientnetB7, Inception V3, MobilenetV2, and Resnet50 V2, as well as 2 distance metrics methods, namely Euclidean and Manhattan. Based on the experiment, the average with 2 measuring distances is more than 90%. The CNN model has the highest retrieval accuracy in Inception V3 with an average of 97.2% MobilenetV2 with an average of 92.9%, Resnet50 V2 with an average of 95.5%, and EfficientNetB7 with an average of 96%. In the skin animal image retrieval accuracy which is more than 70%, the highest number on CNN model Inception V2 is 41 images and EfficientNetb7 is 40 images with Manhattan Distance. Some animal skin image patterns have a retrieval accuracy of up to 100% using 4 CNN models, but some patterns have an accuracy below 50%, so this is part of the continuation of research on animal skin patterns. Keywords-- Manhattan, Euclidean, CNN, EfficientnetB7, InceptionV3, Mobilenet V2, Resnet50 V2

Large Scale Hotel Resource Retrieval Algorithm Based on Characteristic Threshold Extraction

At present, the hotel resource retrieval algorithm has the problem of low retrieval efficiency, low accuracy, low security and high energy consumption, and this study proposes a large scale hotel resource retrieval algorithm based on characteristic threshold extraction. In the large-scale hotel resource data, the mass sequence is decomposed into periodic component, trend component, random error component and burst component. Different components are extracted, the singular point detection is realized by the extraction results, and the abnormal data in the hotel resource data are obtained. Based on the attribute of hotel resource data, the data similarity is processed with variable window, the total similarity of data is obtained, and the abnormal detection of redundant resource data is realized. The abnormal data detection results and redundant data detection results are substituted into the space-time filter, and the data processing is completed. The retrieval problem is identified, and the data processing results are replaced in the hotel resource retrieval based on the characteristic threshold extraction to achieve the normalization of data source and rule knowledge. The characteristic threshold and retrieval strategy are determined, and data fusion reasoning is carried out. After repeated iteration, effective solutions are obtained. The effective solution is fused to get the best retrieval result. Experimental results showed that the algorithm has higher retrieval accuracy, efficiency and security coefficient, and the average search energy consumption is 56n J/bit.

Land Water Vapor Retrieval for AMSR2 Using a Deep Learning Method

In precipitable water vapor (PWV) retrieval, results with high spatial coverage but low temporal resolution can be achieved through satellite-borne sensors, such as the Advanced Microwave Scanning Radiometer 2 (AMSR2). Conversely, the ground-based global navigation satellite system (GNSS) can provide PWV with high temporal resolution and high precision but low spatial coverage. To combine the advantages of these two technologies, we introduce a backpropagation neural network (BPNN) to realize PWV retrieval from AMSR2 with ground-based GNSS data. We first detect the optimal configuration for the BPNN. Then, based on the results of the retrieval accuracy from different types of orbits, we find that the descending (De) orbit has the highest retrieval accuracy, with a root-mean-square error (RMSE) of 3.25 mm. Afterward, the influence of brightness temperature (Tb) data at different frequencies on PWV retrieval is analyzed. The results of GNSS- PWV verification indicate that the 18 +23-GHz frequency combination has the highest PWV retrieval accuracy, and the mean RMSE of all 82 test stations distributed globally can reach 3.53 mm. We also analyze the influence of differently located stations on retrieval accuracy, and the results show that the accuracy of high-latitude and polar regions is remarkably higher than that of other areas but with a lower relative error. Finally, we use radiosonde data as another external verification method to assess PWV retrieval accuracy. The results reveal that RMSE can reach 3.87 mm. Through a BPNN approach, we have creatively realized PWV retrieval from AMSR2 using ground-based GNSS data on a global scale.

Evaluation of Precipitable Water Vapor Product From MODIS and MERSI-II NIR Channels Using Ground- Based GPS Measurements Over Australia

We performed a thorough validation of the precipitable water vapor (PWV) products from near-infrared bands of very similar instruments, i.e., Moderate Resolution Imaging Spectroradiometer (MODIS) and advanced Medium Resolution Spectral Imager (MERSI-II). The PWV products validated are those derived from MODIS/Aqua, MODIS/Terra, and MERSI-II/FY-3D sensors. The PWV data from global positioning system (GPS) in 453 in situ sites situated in the interior and coastal areas of Australia from June 1, 2019 to May 31, 2020 were utilized as reference values. The accuracy of the satellite PWV products was studied under different weather conditions. The evaluation results show that all the three PWV products did not provide good quality water vapor observations in the presence of clouds, with a correlation below 0.2 and a root-mean-square error (RMSE) above 10 mm. Under confident clear sky conditions, MERSI-II/FY-3D instrument had the highest retrieval accuracy with an RMSE of 2.801 mm, but had the worst correlation with reference GPS PWV ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.841). On the contrary, MODIS/Terra instrument had the lowest retrieval accuracy (RMSE = 4.903 mm), but had the best correlation with a correlation of 0.903. Both MODIS/Aqua and MODIS/Terra instruments tended to overestimate PWV value, whereas the MERSI-II/FY-3D instrument tended to underestimate PWV value. All PWV data records showed better retrieval accuracy with higher correlation and lower RMSE under the dry condition than under the wet condition. The performance analysis of all the three satellite PWV products was also studied per day, per station, per season, and per land-surface type in this article.

AI-based multidisciplinary framework to assess the impact of gamified video-based learning through schema and emotion analysis

As a natural and continual process, the initial learning stages encompass mastering and recalling basic facts. The process proves effective with the integration of new information with pre-existing knowledge characterised as schema to facilitate memory encoding. Additionally, emotions also have the ability to modulate human cognition in terms of learning and memory. The recent advent of gamification in e-learning, which has garnered much scholarly and industrial interest, necessitates a thorough examination between video-based learning and its subsequent implications on schema, emotions, and gamification. The current multidisciplinary research triangulated cognitive psychology, affective science, and education technology with artificial intelligence for evaluating digital learning pedagogy based on memory retrieval accuracy, response time, and emotional valence. This three-way (2 x 2 x 2) mixed factorial experiment design with repeated measures entailed 64 healthy young adult volunteers ( n = 64) with 32 in the schema congruent group and 32 in the schema incongruent group. Additionally, 27 (42%) of the volunteers were males, while 37 (58%) were females with an age range between 20 and 39 years old (mean age 27.78 years, SD = 4.77 years). The findings demonstrate that the schema congruent group attained a statistically significant and higher retrieval accuracy ( p < .001). The delayed recall response time was faster than its immediate recall counterpart ( p < .001). Overall, the gamified learning mode depicted more positive emotions compared to non-gamified learning, although both groups primarily portrayed more negative emotions ( p = .05). The synthesis of current research aimed to recommend an AI-based multidisciplinary framework to assess the impact on adult learners in terms of schema and evaluate their emotions in experiencing gamified or non-gamified video materials as a learning medium. The implications expedited from this research offer valuable insights for diverse stakeholders engaged in the video-based learning ecosystem.

Cooperative Cloud-Edge Feature Extraction Architecture for Mobile Image Retrieval

Mobile image retrieval greatly facilitates our lives and works by providing various retrieval services. The existing mobile image retrieval scheme is based on mobile cloud-edge computing architecture. That is, user equipment captures images and uploads the captured image data to the edge server. After preprocessing these captured image data and extracting features from these image data, the edge server uploads the extracted features to the cloud server. However, the feature extraction on the cloud server is noncooperative with the feature extraction on the edge server which cannot extract features effectively and has a lower image retrieval accuracy. For this, we propose a collaborative cloud-edge feature extraction architecture for mobile image retrieval. The cloud server generates the projection matrix from the image data set with a feature extraction algorithm, and the edge server extracts the feature from the uploaded image with the projection matrix. That is, the cloud server guides the edge server to perform feature extraction. This architecture can effectively extract the image data on the edge server, reduce network load, and save bandwidth. The experimental results indicate that this scheme can upload few features to get high retrieval accuracy and reduce the feature matching time by about 69.5% with similar retrieval accuracy.

Knowledge Modeling of power grid regulation based on reasoning map

To contribute to the intelligence and knowledge of power grid regulation and control operations, this paper presents a method of power grid regulation knowledge modeling based on ELG (Event Logic Graph), which includes an event word extraction based on a predicate-argument model, an event chain extraction and fusion based on event similarity theory, an event generalization based on a soft-pattern algorithm, and an event relationship recognition based on rule pattern matching method and joint constraints. Finally, this paper uses events as nodes and event relationships as directed edges to construct an affair graph stipulated by the power grid regulation and control regulations. The ELG is also called the new generation knowledge graph. But the knowledge graph can only describe the existence of entities and the upper and lower associations between entities. ELG can explain the inheritance, causality between entities and the logic of affair evolution, and the probability of transition between legacy and causality. Therefore, knowledge modeling based on ELG has intelligent advantages. Also, compared with ontology-based knowledge modeling methods, the method proposed in this paper can realize the dynamic representation of control operation knowledge, can express the logic of behavior and logic of operation, and also has higher retrieval accuracy.

A National-Scale 1-km Resolution PM2.5 Estimation Model over Japan Using MAIAC AOD and a Two-Stage Random Forest Model

Satellite-based models for estimating concentrations of particulate matter with an aerodynamic diameter less than 2.5 μm (PM2.5) have seldom been developed in islands with complex topography over the monsoon area, where the transport of PM2.5 is influenced by both the synoptic-scale winds and local-scale circulations compared with the continental regions. We validated Multi-Angle Implementation of Atmospheric Correction (MAIAC) aerosol optical depth (AOD) with ground observations in Japan and developed a 1-km-resolution national-scale model between 2011 and 2016 to estimate daily PM2.5 concentrations. A two-stage random forest model integrating MAIAC AOD with meteorological variables and land use data was applied to develop the model. The first-stage random forest model was used to impute the missing AOD values. The second-stage random forest model was then utilised to estimate ground PM2.5 concentrations. Ten-fold cross-validation was performed to evaluate the model performance. There was good consistency between MAIAC AOD and ground truth in Japan (correlation coefficient = 0.82 and 74.62% of data falling within the expected error). For model training, the model showed a training coefficient of determination (R2) of 0.98 and a root mean square error (RMSE) of 1.22 μg/m3. For the 10-fold cross-validation, the cross-validation R2 and RMSE of the model were 0.86 and 3.02 μg/m3, respectively. A subsite validation was used to validate the model at the grids overlapping with the AERONET sites, and the model performance was excellent at these sites with a validation R2 (RMSE) of 0.94 (1.78 μg/m3). Additionally, the model performance increased as increased AOD coverage. The top-ten important predictors for estimating ground PM2.5 concentrations were day of the year, temperature, AOD, relative humidity, 10-m-height zonal wind, 10-m-height meridional wind, boundary layer height, precipitation, surface pressure, and population density. MAIAC AOD showed high retrieval accuracy in Japan. The performance of the satellite-based model was excellent, which showed that PM2.5 estimates derived from the model were reliable and accurate. These estimates can be used to assess both the short-term and long-term effects of PM2.5 on health outcomes in epidemiological studies.

An investigation on microwave transmissivity at frequencies of 18.7 and 36.5 GHz for diverse forest types during snow season

ABSTRACT Forests have invariably been considered as an obstacle in retrieving land surface parameters from spaceborne passive microwave brightness temperature (TB) observations. For quantifying the effect of forests on microwave signals, several models have been developed. However, these models rarely reveal the dependence of microwave radiation on forest types, which can hardly meet the needs of high-accuracy retrieval of terrestrial parameters in forested regions. A ground-based microwave radiometric observation experiment was designed to investigate the dependence of microwave radiation on frequency, polarization, and forest type. Downward TB at 18.7- and 36.5-GHz for horizontal- and vertical-polarization from the forest canopy was measured at 14 sample plots in Northeast China, along with snowpack and forest structural parameters. By providing fits to experimental data, new empirical transmissivity models for three forest types were developed, as a function of woody stem volume and depending on the frequency/polarization. The proposed models give diverse asymptotic transmissivity saturation levels and the corresponding saturation point of woody stem volume for different forest types. Root-mean-square error results between TB simulations and Advanced Microwave Scanning Radiometer-2 observations are approximately 3–6 K. This study provides an experimental and theoretical reference for further development of inversion models for snow parameters in forested areas.

Retrieval of behavior trees using map-and-reduce technique

There has been an increased interest in the creation of AI social agents who possess complex behaviors that allow them to perform social interactions. Behavior trees provide a plan model execution that has been widely used to build complex behaviors for AI social agents. Behavior trees can be represented in the form of a memory structure known as cognitive scripts, which would allow them to evolve through further development over multiple exposure to repeated enactment of a particular behavior or similar ones. Behavior trees that share the same context will then be able to learn from each other resulting in new behavior trees with richer experience. The main challenge appears in the expensive cost of retrieving contextually similar behavior trees (scripts) from a repertoire of scripts to allow for that learning process to occur. This paper introduces a novel application of map-and-reduce technique to retrieve cognitive with low computational time and memory allocation. The paper focuses on the design of a corpus of cognitive scripts, as a knowledge engineering key challenge, and the application of map-and-reduce with semantic information to retrieve contextually similar cognitive scripts. The results are compared to other techniques used to retrieve cognitive scripts in the literature, such as Pharaoh which uses the least common parent (LCP) technique in its core. The results show that the map-and-reduce technique can be successfully used to retrieve cognitive scripts with high retrieval accuracy of 92.6%, in addition to being cost effective.

A Comparative Analysis of Retrieval Algorithms of Land Surface Temperature from Landsat-8 Data: A Case Study of Shanghai, China.

In the trend of global warming and urbanization, frequent extreme weather has a severe impact on the lives of citizens. Land Surface Temperature (LST) is an essential climate variable and a vital parameter for land surface processes at local and global scales. Retrieving LST from global, regional, and city-scale thermal infrared remote sensing data has unparalleled advantages and is one of the most common methods used to study urban heat island effects. Different algorithms have been developed for retrieving LST using satellite imagery, such as the Radiative Transfer Equation (RTE), Mono-Window Algorithm (MWA), Split-Window Algorithm (SWA), and Single-Channel Algorithm (SCA). A case study was performed in Shanghai to evaluate these existing algorithms in the retrieval of LST from Landsat-8 images. To evaluate the estimated LST accurately, measured data from meteorological stations and the MOD11A2 product were used for validation. The results showed that the four algorithms could achieve good results in retrieving LST, and the LST retrieval results were generally consistent within a spatial scale. SWA is more suitable for retrieving LST in Shanghai during the summer, a season when the temperature and the humidity are both very high in Shanghai. Highest retrieval accuracy could be seen in cultivated land, vegetation, wetland, and water body. SWA was more sensitive to the error caused by land surface emissivity (LSE). In low temperature and a dry winter, RTE, SWA, and SCA are relatively more reliable. Both RTE and SCA were sensitive to the error caused by atmospheric water vapor content. These results can provide a reasonable reference for the selection of LST retrieval algorithms for different periods in Shanghai.

A Robust Atmospheric Correction Procedure for Determination of Spectral Reflectance of Terrestrial Surfaces from Satellite Spectral Measurements

In this work, we propose simple and robust technique for the retrieval of underlying surface spectral reflectance using spaceborne observations. It can be used to process both multispectral moderate resolution satellite data and also multi-zone high spatial resolution data. The technique can work automatically for different types of land surfaces without using huge databases accumulated in advance. The new cloud discrimination and retrieval of the water vapor content in atmosphere procedures are presented. The key point of the proposed atmospheric correction technique is the suggested single-wavelength method for determining the atmospheric aerosol optical thickness without reference to a specific type of underlying surface spectrum. The retrievals of spectral reflectance for various land surfaces with developed technique, performed using computer simulation and experimental data, have demonstrated a high retrieval accuracy.

Relevance Feedback For Image Retrieval Using Transfer Learning and Improved MQHOA

Image retrieval is a challenging technology in multimedia applications where meeting the users’ subjective retrieval needs while achieving high retrieval performance is insufficient for existing methods. In this work, a related feedback image retrieval algorithm based on deep learning and optimization algorithm (CAMQHOA-RF) is proposed. Transfer learning based on the deep convolutional neural network is applied to extract deeper image features to reduce the semantic gap. The multi-scale quantum harmonic oscillator algorithm improved by the idea of “aggregation” is introduced to search the feature space effectively. The covariance matrix is used to strengthen the relationship between feature points at different scales to guide feature points to approach ideal query points faster. Moreover, the query point is reselected based on the feedback information to explore more potential users’ interest areas. Experiments have shown that compared with other algorithms, the proposed algorithm has fewer parameters that need to be set, but higher retrieval accuracy, faster retrieval speed, and stronger robustness are obtained, which can meet users better.

A secure heuristic semantic searching scheme with blockchain-based verification

Secure semantic searching provides privacy-preserving information retrieval for arbitrary queries semantically related to keywords. However, current verification mechanisms cannot verify the correctness of search results for semantically related queries in mutual distrust scenarios. Moreover, most secure semantic searching schemes still perform exact matching on ciphertext after query expansion, then directly accumulate the weights of the matched keywords as similarity measurements to rank the documents, resulting in the search results with unsatisfied ranking. In this paper, we propose a secure heuristic semantic searching scheme, in which a privacy-preserving word nonlinear matching (PPWNM) method is developed, and a blockchain-based verification is designed to obtain trustworthy search results with high retrieval accuracy. Specifically, we describe several retrieval heuristics and formulate them as a word nonlinear matching problem, then transform it to the PPWNM problem for calculating the similarity measurement between the query and document so that the highly accurate ranked results can be obtained. Also we design a blockchain-based verification mechanism in which the blockchain nodes utilize the proofs generated during the matching process to verify the correctness of the search results and reach a consensus to ensure the trustworthy results and fair payment between the participants in a mutual distrust model. Our security analysis and experimental results show that the proposed scheme is secure and has higher accuracy compared with the existing other schemes.