Content Retrieval Research Articles

Towards Generative Search and Recommendation: A keynote at RecSys 2023

The emergence of large language models (LLM's), especially ChatGPT, has for the first time make AI known to almost everyone and affected every facet of our society. The LLMs have the potential to revolutionize the ways we seek and consume information. This has stemmed the recent trends in both academia and industry to develop LLM-based generative AI systems for various applications with enhanced capabilities. One such systems is the generative search and recommender system, which is capable of performing content retrieval, content repurposing, content creation and their integration to meet users' information needs. However, before such systems can be widely used and accepted, we need to address several challenges. The primary challenge is the trust and safety in the generated content as we expect the LLM's to make mistakes with hallucination. This is because of the quality of data being used for their training is often erroneous and biased. The other challenges in the search and recommendation domain include: how to teach the system to be pro-active in anticipating the needs of users and in directing the conversation towards a fruitful direction; as well as the integration of retrieved and generated content. This keynote presented a generative information seeking paradigm, and discuss key research towards a trustable generative system for search and recommendation. Date : 21 September 2023.

Retrieving decametric-resolution leaf chlorophyll content from GF-6 WFV by assessing the applicability of red-edge vegetation indices

Accurate and timely retrieval of leaf chlorophyll content (LCC) at fine spatial scales is crucial for monitoring plants' photosynthetic capacity and nutritional status. The wide field-of-view camera (WFV) onboard Gaofen-6 (GF-6) satellite has two red-edge bands at 16-meter spatial resolution, providing an opportunity to derive decametric-resolution LCC. However, the potential of GF-6 WFV data in LCC estimation has yet to be studied. Vegetation indices (VIs) are widely used in LCC estimation due to their simplicity and efficiency, and the accuracy of LCC inversion can be improved with red-edge VIs. This study focused on exploring the retrieval of LCC with GF-6 WFV using the VI-based method by assessing the applicability of nine red-edge VIs derived from GF-6 WFV. Our results revealed that (1) the newly developed canopy structure-insensitive CSI (chlorophyll sensitive index) achieved comparable accuracy to the physically-based method and reduced the limitation of the VI-based method being species- and time-specific. Validation results showed that the root-mean-square error (RMSE) of the estimated LCC from the CSI-based model was 7.91–11.38 μg/cm2 for four plant functional types (similar accuracy to Croft et al. (2020) based on the lookup table method with RMSE of 9.25–13.18 μg/cm2). The accuracy of the CSI-based model remained stable across species and growth conditions, indicating its feasibility for large-scale applications. (2) GF-6 WFV provided accurate LCC estimates and well-captured the spatial and seasonal variations in LCC. GF-6 LCC maps showed comparable accuracy (R2 = 0.53, RMSE = 9.78 μg/cm2) and similar spatiotemporal distribution to the 30-m Sentinel-2 LCC products. Additionally, GF-6 LCC maps outperformed Sentinel-2 LCC products regarding observation frequency and spatial continuity. Our study highlights the potential of GF-6 WFV in retrieving decametric-resolution LCC and provides further support for applications of GF-6 WFV data in precision agriculture and forestry.

Retrieval of chlorophyll content for vegetation communities under different inundation frequencies using UAV images and field measurements

Karst wetlands are an important type of wetland among the global karst areas that suffer from serious ecosystem degradation. Chlorophyll content, an important parameter of vegetation functional traits, can indicate the photosynthetic activity, mutation and stress level, and nutritional state of vegetation. However, quantitative estimation of canopy chlorophyll content (CCC) of calcium-loving and alkali-resistant vegetation in karst wetlands using remote sensing technology for clarifying the relationship between canopy spectral reflectance and the CCC of vegetation communities remains a challenge. In this study, we propose a retrieval approach for estimating the CCC of seven typical vegetation communities in the largest karst wetland in China, using a combination of parameter/non– parameter estimation model, and unmanned aerial vehicle (UAV) multispectral images. We further evaluated the sensitivity of various spectral indices to the CCC of the vegetation communities and explored the influence of image feature distribution on CCC estimation accuracy. We revealed the variation patterns of leaf chlorophyll content (LCC) and CCC under different inundation frequencies in the karst wetland. We confirmed that the UAV-based inversion method proposed in this study achieved the CCC estimation of seven vegetation communities (R2 = 0.57 ∼ 0.93, RMSE = 8.80 SPAD ∼ 60.68 SPAD), and obtained high estimation accuracy (R2 = 0.77, RMSE = 22.04 SPAD) of Cladium chinense Nees (CCN, endemic vegetation in karst wetland). We found that the optimal reflectance ranges for estimating the CCC of the seven vegetation communities were 714 nm ∼ 746 nm and 814 nm ∼ 866 nm, and that the ChIrededge and MSR indices negatively influenced the CCC estimation of CCN and Carex baccans Nees (CBN), respectively. We demonstrated that vegetation communities with well-defined canopy structure (Triadica sebifera (Linnaeus) Small) achieved higher estimation accuracy with moderate uncertainty (ε = 10.80 % ∼ 15.49 %) than the herbaceous vegetation communities (CCN, CBN, and Cynodon dactylon (L.) Pers.). The bias of the CCC estimation based on linear regression and partial least squares regression models in different vegetation communities ranged from 1.61 % to 49.41 %. We confirmed that the magnitude of variations in LCC and CCC for these vegetation communities presented a decreasing pattern from that in high-frequency inundation communities to that in the low-frequency ones.

Bridging Qualitative Data Silos: The Potential of Reusing Codings Through Machine Learning Based Cross-Study Code Linking

For qualitative data analysis (QDA), researchers assign codes to text segments to arrange the information into topics or concepts. These annotations facilitate information retrieval and the identification of emerging patterns in unstructured data. However, this metadata is typically not published or reused after the research. Subsequent studies with similar research questions require a new definition of codes and do not benefit from other analysts’ experience. Machine learning (ML) based classification seeded with such data remains a challenging task due to the ambiguity of code definitions and the inherent subjectivity of the exercise. Previous attempts to support QDA using ML rely on linear models and only examined individual datasets that were either smaller or coded specifically for this purpose. However, we show that modern approaches effectively capture at least part of the codes’ semantics and may generalize to multiple studies. We analyze the performance of multiple classifiers across three large real-world datasets. Furthermore, we propose an ML-based approach to identify semantic relations of codes in different studies to show thematic faceting, enhance retrieval of related content, or bootstrap the coding process. These are encouraging results that suggest how analysts might benefit from prior interpretation efforts, potentially yielding new insights into qualitative data.

Survey on the scheme evaluation, opportunities and challenges of software defined‐information centric network

AbstractAs a promising architecture of next‐generation network, software defined‐information centric network (SD‐ICN) inherits the advantages of software defined network (SDN) and information‐centric network (ICN) to enable flexible and fast content retrieval, especially in the current era of artificial intelligence. However, the existing researches mainly focus on a single respective in this field, which motivates in comprehensively providing a forward‐looking guidance and development direction for scholars and engineers. To this end, the latest developments of SD‐ICN is presented. First, the widely‐accepted concepts and impacts on traditional networks are introduced. Second, the shortcomings of SDN and ICN over conventional networks are respectively analyzed to illustrate the necessity of SD‐ICN. Third, based on extensive analysis and deep deliberation, a methodical taxonomy for existing combination studies is proposed. They are divided into SDN over ICN, ICN over SDN, and mutual immersive pattern. Fourth, the performances of three integration categories are compared and the limitations of related works are highlighted. Fifth, the maturity index from six development indicators are evaluated. Further, the maturity and practicality of these schemes are generalized. Based on the above studies and comparisons, the lessons learned by SDN and ICN developments are concluded. Finally, future research directions and opportunities are discussed for the readers.

Pseudorange characteristics of soil-transmitted GNSS signals and their application in soil moisture content retrieval

Pseudorange characteristics of soil-transmitted GNSS signals and their application in soil moisture content retrieval

Estimation of Soil Salt Content at Different Depths Using UAV Multi-Spectral Remote Sensing Combined with Machine Learning Algorithms

Soil salinization seriously affects the sustainable development of agricultural production; thus, the timely, efficient, and accurate estimation of soil salt content (SSC) has important research significance. In this study, the feasibility of soil salt content retrieval using machine learning models was explored based on a UAV (unmanned aerial vehicle) multi-spectral remote sensing platform. First, two variable screening methods (Pearson correlation analysis and Grey relational analysis) are used to screen the characteristic importance of 20 commonly used spectral indices. Then, the sensitive spectral variables were divided into a vegetation index group, a salt index group, and a combination variable group, which represent the model. To estimate SSC information for soil depths of 0–20 cm and 20–40 cm, three machine learning regression models were constructed: Support Vector Machine (SVM), Random Forest (RF), and Backpropagation Neural Network (BPNN). Finally, the salt distribution map for a 0–20 cm soil depth was drawn based on the best estimation model. The results of experiments show that GRA is better than PCA in improving the accuracy of the estimation model, and the combination variable group containing soil moisture information performs best. The three machine learning models have achieved good prediction effects to some extent. The accuracy and stability of the model are considered comprehensively, the prediction effect of 0–20 cm is higher than that of 20–40 cm, and the validation set coefficient of determination (R2), Root-Mean-Square-Error (RMSE), and Mean Absolute Error (MAE) of the best inversion model are 0.775, 0.055, and 0.038, and the soil salt spatial map based on the optimal estimation model can reflect the salinization distribution in the study area. Therefore, this study shows that a UAV multi-spectral remote sensing platform combined with machine learning models can better monitor farmland soil salt content.

Retrieval of leaf chlorophyll content using drone imagery and fusion with Sentinel-2 data

Leaf chlorophyll is an important dynamic biochemical parameter to assess crop phenology, health stress, and yield. Optical remote sensing data is a widely used technique for the estimation of vegetation leaf chlorophyll. Drone technology is a promising solution for high-resolution monitoring of the leaf chlorophyll content and employed in this study. Image fusion was also accomplished to provide high-resolution multi-spectral images by combining Sentinel-2A and drone imagery. Support vector machine regression was adopted to determine the best vegetation indices for the retrieval of leaf chlorophyll content. Random forest regression and support vector machine regression algorithms were adopted to develop the best models for chlorophyll retrieval. The 34 models derived from drone data and the 46 models derived from fusion data were evaluated for chlorophyll retrieval. It was found that the best support vector machine model, based on anthocyanin content index and enhanced vegetation index (M31), had the best correlation coefficient (r = 0.732) and root mean square error (1.93). The random forest regression models M32 (based on chlorophyll vegetation index and visible atmospherically resistant index) and M16 (based on canopy chlorophyll content index and visible atmospherically resistant index) for drone data were found to be the best, which performed equally well in terms of correlation coefficients (r = 0.77) and root mean square error (1.51). The multiple vegetation indices of drone-based leaf chlorophyll content random forest regression models (M16 and M32) provided higher performance than single vegetation indices-based linear (simple linear fit) and nonlinear leaf chlorophyll content models. Out of 46 models of fusion products, the random forest regression M14 (green normalized difference vegetation index with narrow near infrared band and visible atmospherically resistant index) offered the best performance for leaf chlorophyll content retrieval, as can be observed through a comparison of correlation (r = 0.77) and RMSE (=1.71) values. It was also investigated that inversion performances (r ∼ 0.8 and RMSD ∼ 1.4 with standard deviation ∼ 2.6) of fusion data-based models in cross-applicability and transferability were found to be suitable for large-scale inversion of Sentinel-2A data products.

AI-BASED VIDEO SUMMARIZATION FOR EFFICIENT CONTENT RETRIEVAL

The explosive growth of video data poses a significant challenge in retrieving relevant content swiftly. Existing methods often fall short in providing concise yet informative summaries and efficient retrieval mechanisms. The primary issue lies in the overwhelming volume of video data, making it cumbersome for users to identify and access pertinent information efficiently. Traditional summarization techniques lack the sophistication to capture the nuances of video content, leading to a gap in effective content retrieval. Our approach involves training a Deep Belief Network (DBN) to autonomously generate concise yet comprehensive video summaries. Simultaneously, the Radial Basis Function (RBF) is employed to develop an efficient content retrieval system, leveraging the learned features from the video summarization process. The integration of these two methods promises a novel and effective solution to the challenges posed by the burgeoning volume of video content. Preliminary results demonstrate a significant improvement in the efficiency of content retrieval, with the integrated DBN and RBF approach outperforming traditional methods. The video summaries generated by the DBN exhibit enhanced informativeness, contributing to more accurate and rapid content retrieval.

Mind mapping in recalling and retrieving core contents in anesthesia technology students

BACKGROUND: Human memory represents the ability to recall and retrieve information. Learning methods have a great impact on remembering and retrieving material. This study was conducted to investigate the effect of the mind map learning method on the amount of recall and retrieval of the core contents of anesthesia technology students. MATERIALS AND METHODS: In this quasi-experimental study, 55 undergraduate students of anesthesia technology were included in the study by census method. In both groups, teaching was done in the same way (lecture method) by a single professor. The intervention group was randomly divided into six groups. After training to work with a mind map in this group, the teaching of lessons started with the lecture method, and then the mind map was drawn by the group members. In the control group, the students followed the lecture method and note-taking. Finally, three tests were taken. Test 1 (before the start of the training), test 2 (immediately after the training (scheduled test)), and test 3 (after 2 weeks (surprise test)). RESULTS: The data obtained from the test scores were analyzed by paired t-test, independent t-test, and analysis of covariance. The analysis showed that the mind map method significantly improved the short-term (P < 0.001) and long-term memory of anesthesia technology students (P < 0.001) compared to the traditional method. Examining the effect of gender on long-term memory showed that in the intervention group, female samples had an average score of 1 more than male samples (P = 0.044). DISCUSSION: It seems that mind mapping improves short-term and long-term memory in anesthesia technology students who play a critical role in saving the lives of patients and this issue improves the quality of services in the field of health.

Dynamic and Energy Efficient Cache Scheduling Framework for IoMT over ICN

The Internet of Medical Things (IoMT) is the network of medical devices, hardware infrastructure, and software applications used to connect the healthcare information technology. Massive traffic growth and user expectations cause challenges in the current exhausting models of IoMT data. To reduce the IoMT traffic, Information Centric Network (ICN) is a suitable technique. ICN uses persistent naming multicast communication that reduces the response time. ICN in IoMT provides a promising feature to reduce the overhead due to the distribution of commonly accessed contents. Some parameters such as energy consumption, communication cost, etc., influence the performance of sensors in the IoMT network. Excessive and unbalanced energy consumption degrades the network performance and lifetime. This article presents a framework called Dynamic Cache Scheme (DCS) that implements energy-efficient cache scheduling in IoMT over ICN to reduce network traffic. The proposed framework establishes a balance between the multi-hop traffic and data item freshness. The technique improves the freshness of data; thus, updated data are provided to the end-users via the effective utilization of caching in IoMT. The proposed framework is tested on important parameters, i.e., cache-hit-ratio, stretch, and content retrieval latency. The results obtained are compared with the state-of-the-art models. Results’ analysis shows that the proposed framework outperforms the compared models in terms of cache-hit-ratio, stretch, and content retrieval latency by 59.42%, 32.66%, and 18.8%, respectively. In the future, it is intended to explore the applicability of DCS in more scenarios and optimize further.

Research on the development of literature practice teaching for college students in the context of computer technology

Abstract In this paper, intelligent information retrieval of relevant literary content through collaborative filtering algorithms in Internet technology is used to analyze the similarity between users and, thus, to select similar items in a weighted ranking. The Jacobian formula is used to calculate the common feature values and thus predict the user’s unrated item ratings. Word normalization is performed through information extraction of text keywords to effectively abstract the dimensionality of text features. The results showed that the influence of learning based on language literacy was 0.697, the preference for the literary genre of prose was as high as 49.4%, and students could accept infiltration teaching. It indicates that making full use of Internet technology means can break the time and space limitations in the traditional teaching of literature, provide effective support for students’ interactive communication, expand students’ cultural horizons, and improve their humanistic literacy.

Analysis of the effect of user-understanding-based retrieval system improvement on the application of intelligent archive management

Abstract As a precipitation and accumulation of history, archives management has gradually tended to be digitalized and informalized with the continuous updating and development of modern technology. In this paper, we first study the retrieval system and focus on the Boolean model, vector model, and probabilistic model in information retrieval technology. The matching relationship between documents and queries is detected from the document set for the user’s query, and a relevance retrieval system based on user understanding is proposed to solve the matching problem. The amount of information that needs to be retrieved is growing exponentially, and how a user perceives the information is crucial to the process. Then, in order to solve the problem of insufficient retrieval efficiency caused by the explosive growth of wisdom files, the retrieval system is creatively optimized on the basis of the ant colony algorithm, which effectively improves the efficiency of wisdom file management. The efficiency of the optimized retrieval system is verified and analyzed in an experimental simulation environment. The findings demonstrate that when the amount of archives rises, the retrieval effectiveness of the improved ant colony algorithm described in this study marginally improves, but in 10~35s. As the inventory of the Smart Archives increases, the content retrieval of the archives will become more and more frequent. This study improves retrieval efficiency and serves as a good demonstration for the construction of archival management information technology.

NDN-Based Coded Caching Strategy for Satellite Networks

To solve the transmission correlation issue arising from traditional named data networking (NDN) caching during segmenting contents, where users must obtain data blocks related to the requested file blocks to recover the source data, in this paper, we propose an NDN-based coded caching strategy for low Earth orbit (LEO) satellite networks, using coding operations to remove the transmission correlation of data. To achieve efficient content distribution, the satellite node’s coded package is strategically placed with the objectives of minimizing the backhaul link load and content acquisition latency. The optimization problem is solved by using a multi-colony ant colony algorithm, enabling fast content retrieval. We designed a cluster cooperative distribution mechanism to simplify the satellite network’s management and reduce the load of satellite links for content distribution when covered by multiple satellite nodes. Finally, we compare the multi-objective coded caching (CCMO) introduced in this paper with the most popular (MP) strategy and the random-based coded caching (CCR) strategy; the average cache hit ratio, backhaul link load, and content acquisition delay are compared and analyzed. The results show that CCMO performs better.

A SEMWORD based Semantic Secure Content Retrieval System in E-learning

A SEMWORD based Semantic Secure Content Retrieval System in E-learning

Erratum to: Retrieval of HDO Relative Content in the Atmosphere from Simultaneous GOSAT-2 Measurements in the Thermal and Near-IR

Erratum to: Retrieval of HDO Relative Content in the Atmosphere from Simultaneous GOSAT-2 Measurements in the Thermal and Near-IR

Retrieval of Soil Heavy Metal Content for Environment Monitoring in Mining Area via Transfer Learning

Monitoring environmental pollution sources is an ongoing issue that must be addressed to reduce risks to public health, food safety, and the environment. However, retrieving topsoil heavy metal content at a low cost for environmental monitoring in mining areas is challenging. Therefore, this study proposes a network model based on transfer learning theory and a back propagation (BP) network optimized by a genetic algorithm (GA), taking the Daxigou mining area in Shaanxi Province, China, as a case study. Firstly, visible and near-infrared spectrum data from Landsat8 satellite images, digital elevation models, and geochemical data from field-collected soil samples were used to extract environmental factor candidates indicating the content and spatial distribution of certain heavy metals, including copper (Cu) and lead (Pb). Secondly, each element was correlated with environmental factors and a multicollinearity test was performed to determine the optimal factor set. Then, the BP network optimized by GA was pre-trained with sample data collected in 2017 and retrained with minimal sample data from 2019 using the parameter transfer learning method, allowing spatial distribution mapping of the Cu and Pb content in topsoil of the Daxigou mining area in 2019. From the validation results using field-collected data, the root mean square error (RMSE) and mean relative error (MRE) values using the proposed model, respectively, reduced by 4.688 mg/kg and 1.533 mg/kg for Cu and reduced by 1.586 mg/kg and 1.232 mg/kg for Pb compared to the traditional GA-BP model. Thus, conclusions can be drawn that our proposed Tr-GA-BP network performs well, requiring 16 training samples collected in 2019. In addition, the content of Cu is the highest; Pb is the second highest in the study area. Both of them were spatially distributed mainly in the exploitation, slag stacking, roadside, etc., consistent with field investigation results.

A high-performance speech BioHashing retrieval algorithm based on audio segmentation

As one of the research hotspots in the field of speech recognition, content-based speech retrieval algorithms can detect speech information with the same content features, which improves computer intelligence while reducing labor costs, and thus have been widely used. Although most of the current speech content retrieval algorithms can guarantee excellent retrieval performance for small-scale speech retrieval work, the performance of the above algorithms is greatly reduced under the constraints of large speech data storage space and high content redundancy. In order to solve the above problems, a high-performance speech BioHashing retrieval algorithm based on audio segmentation is proposed in this paper. The algorithm is divided into an offline pre-processing phase and an online retrieval phase, The offline pre-processing stage converts the speech data into BioHashing sequences with speech content characteristics. In this process, first of all, the Power-Normalized Cepstral Coefficients (PNCC) features of the speech data are extracted and biometric templates with single mapping keys are constructed according to the PNCC features, obtaining BioHashing sequences. Then, slice the original speeches into short-time audio segments according to the proposed audio segmentation algorithm, and the hash reconstruction operation is performed on the BioHashing sequences to obtain the reconstructed Hashing sequences for online retrieval. The online search phase responds to the users' query requests, just find the hash index that matches the query hash sequence from the BioHashing index table, and will the standardized editing distance (SED) to the closest 1 value corresponding to the hash index as the retrieval result back to the user. The experimental results show that the reconstructed hash sequences obtained after removing the silent redundant segments have better robustness and discrimination. Moreover, the algorithm achieves 100% retrieval accuracy for the original speech clips, and the average retrieval time is only 0.0157 s, which shows that the algorithm has good retrieval performance and can meet the needs of speech retrieval in various environments.

Query-by-Example Spoken Term Detection for Zero-Resource Languages Using Heuristic Search

Query-by-Example spoken content retrieval is a demanding and challenging task when a large volume of spoken content is piled up in the repositories without annotation. In the absence of annotation, spoken content retrieval is achieved by capturing the similarities between the query and spoken terms from the acoustic feature representation itself. Dynamic Time Warping (DTW) centric techniques identify the optimal alignment between the acoustic feature representations and capture the similarities between query and spoken terms. Despite feasibility, the DTW-centric techniques produce a lot of false alarms due to the variabilities that exist in natural speech and degrade the performance. In the proposed approach, the variability challenges are addressed in two stages. At first, the speaker-independent acoustic feature representation was obtained from the deep convolutional neural networks that reduce the speaker variabilities. In the second stage, the similarities between the query and spoken term were captured using the heuristic search method. The proposed approach was compared with other state-of-the-art methods using Microsoft Low-Resource Language speech corpus. A 3% improvement and 32% reduction in the hit and false alarm ratio were achieved across languages.

Juliet-PUF: Enhancing the Security of IoT-Based SRAM-PUFs Using the Remanence Decay Effect

The cloud-based Internet of Things (IoT) enables the creation of innovative computer applications based on sensing, analyzing, and controlling the physical world. IoT deployments, however, are at a particular risk of counterfeiting, through which an adversary can corrupt the entire ecosystem. Therefore, entity authentication of edge devices is considered an essential part of the security of IoT systems. This research addresses the challenge of generating a unique ID in IoT devices. Unique IDs allow the IoT system maker to identify each edge device, and to ensure that only genuine devices can upload data to the cloud. Traditional ID mechanisms are not feasible in IoT, due to the edge device’s constrained runtime environment, or the additional costs and the deployment difficulties that they introduce. In this work, we present JULIET-PUF, a novel PUF-based method for IoT identification, which relies on SRAM content retrieval after power glitches with time differences. Our scheme comes with no added hardware cost on the edge device. We evaluate JULIET-PUF using a dataset of 24 units of a popular commercial IoT device, and show that it is on average 95.58 times more secure than the common use of SRAM-PUF.