Relevance Feedback Research Articles

Application of entertainment interactive robot based on speech recognition in English artificial intelligence teaching evaluation and automatic feedback

With the development of intelligent voice and interactive robot technology, new technologies have built a virtual E-learning learning environment that can provide students with an immersive learning experience, making this new learning mode more entertaining. This article investigates the application of entertainment interactive robots based on speech recognition in English artificial intelligence teaching evaluation and automatic feedback. The system has constructed an oral evaluation model based on deep reinforcement learning, which learns the optimal behavioral strategies through interaction with the environment. The model will train through oral conversations with learners to learn how to accurately evaluate oral proficiency and provide relevant feedback. After the construction of the system is completed, the accuracy and efficiency of the system are improved by adjusting the parameters of the model, increasing the diversity of training data, and improving the user interface and interaction mode based on user feedback, making it more friendly and easy to use. The experimental results show that the English oral evaluation and automatic feedback system designed in this paper based on deep reinforcement learning and speech recognition algorithms has high accuracy and efficiency. The system can accurately evaluate learners’ oral proficiency and provide personalized learning suggestions based on individual differences.

An NLP-Based Exploration of Variance in Student Writing and Syntax: Implications for Automated Writing Evaluation

In writing assessment, expert human evaluators ideally judge individual essays with attention to variance among writers’ syntactic patterns. There are many ways to compose text successfully or less successfully. For automated writing evaluation (AWE) systems to provide accurate assessment and relevant feedback, they must be able to consider similar kinds of variance. The current study employed natural language processing (NLP) to explore variance in syntactic complexity and sophistication across clusters characterized in a large corpus (n = 36,207) of middle school and high school argumentative essays. Using NLP tools, k-means clustering, and discriminant function analysis (DFA), we observed that student writers employed four distinct syntactic patterns: (1) familiar and descriptive language, (2) consistently simple noun phrases, (3) variably complex noun phrases, and (4) moderate complexity with less familiar language. Importantly, each pattern spanned the full range of writing quality; there were no syntactic patterns consistently evaluated as “good” or “bad”. These findings support the need for nuanced approaches in automated writing assessment while informing ways that AWE can participate in that process. Future AWE research can and should explore similar variability across other detectable elements of writing (e.g., vocabulary, cohesion, discursive cues, and sentiment) via diverse modeling methods.

3VR: Vice Versa Virtual Reality Algorithm to Track and Map User Experience

The understanding of how users interact with the virtual cultural heritage could provide digital curators valuable insights into user behaviors and also improve the overall user experience through the ability to observe and record interactions of virtual visitors. This article introduces the new User Behavior (UB) tracking algorithm that we developed investigating a salience of the Virtual Reality (VR) panoramic regions. The algorithm extracts the importance of Region of Interest (ROI) determining patterns of the visitors’ virtual movement and interest in combination with statistics of captured browser activity. The input of our algorithm is the virtual online interactive platform (Virtual Museum of the Civic Art Gallery of Ascoli Piceno in Italy) with 81 16,386 × 8,192 pixels panoramic images and several interactive features including maps, thumbnails, and menus. The software engine of the tracking model “Vice Versa” VR operates on inverse functions of all descriptive functions (descriptors), which are assigned particularly to each interactive feature such as viewing multimedia content and observing the panoramic environment. The tracking experiment was performed online and the web virtual museum key study collected behavior information from 171 visitors around the world. Collected data, multimedia and textual content, and the coordinates of the ROIs are then subjected to standard statistics operations to define common patterns of UBs. Thus, we have discovered that the ROIs are mostly mapped onto the artworks and it is possible to obtain patterns about the main interests of users. The developed tool offers a guideline for the panoramic tours design and the potential benefits for museums are to understand the public, verify the effectiveness of choices, and re-shape a cultural offer based on visitors’ needs. Exploiting this kind of user experience, our algorithm ensures relevant feedback during virtual visits and thus paves the way for further development of the recommender system.

ROGER: Ranking-oriented Generative Retrieval

In recent years, various dense retrieval methods have been developed to improve the performance of search engines with a vectorized index. However, these approaches require a large pre-computed index and have limited capacity to memorize all semantics in a document within a single vector. To address these issues, researchers have explored end-to-end generative retrieval models that use a seq-to-seq generative model to directly return identifiers of relevant documents. Although these models have been effective, they are often trained with the maximum likelihood estimation method. It only encourages the model to assign a high probability to the relevant document identifier, ignoring the relevance comparisons of other documents. This may lead to performance degradation in ranking tasks, where the core is to compare the relevance between documents. To address this issue, we propose a ranking-oriented generative retrieval model that incorporates relevance signals in order to better estimate the relative relevance of different documents in ranking tasks. Based upon the analysis of the optimization objectives of dense retrieval and generative retrieval, we propose utilizing dense retrieval to provide relevance feedback for generative retrieval. Under an alternate training framework, the generative retrieval model gradually acquires higher-quality ranking signals to optimize the model. Experimental results show that our approach increasing Recall@1 by 12.9% with respect to the baselines on MS MARCO dataset.

Implicit Adaptation Is Modulated by the Relevance of Feedback.

Given that informative and relevant feedback in the real world is often intertwined with distracting and irrelevant feedback, we asked how the relevancy of visual feedback impacts implicit sensorimotor adaptation. To tackle this question, we presented multiple cursors as visual feedback in a center-out reaching task and varied the task relevance of these cursors. In other words, participants were instructed to hit a target with a specific task-relevant cursor, while ignoring the other cursors. In Experiment 1, we found that reach aftereffects were attenuated by the mere presence of distracting cursors, compared with reach aftereffects in response to a single task-relevant cursor. The degree of attenuation did not depend on the position of the distracting cursors. In Experiment 2, we examined the interaction between task relevance and attention. Participants were asked to adapt to a task-relevant cursor/target pair, while ignoring the task-irrelevant cursor/target pair. Critically, we jittered the location of the relevant and irrelevant target in an uncorrelated manner, allowing us to index attention via how well participants tracked the position of target. We found that participants who were better at tracking the task-relevant target/cursor pair showed greater aftereffects, and interestingly, the same correlation applied to the task-irrelevant target/cursor pair. Together, these results highlight a novel role of task relevancy on modulating implicit adaptation, perhaps by giving greater attention to informative sources of feedback, increasing the saliency of the sensory prediction error.

Small programs, big challenges: Reimagining the evaluation of clinical teaching in genetic counseling.

Within the health professions education system, a significant proportion of teaching and learning occurs in the clinical setting. As such, the need to measure effective teaching for accreditation standards, faculty development, merit pay, academic promotion, and for monitoring the safety of the learning environment has led to numerous universities developing instruments to evaluate teaching effectiveness in this context. To date; however, these instruments typically focus on the student perspective, despite evidence demonstrating that student evaluations of teaching (SETs) lack correlation with learning outcomes and are not a true measure of teaching effectiveness. This issue is further exacerbated in small health professional training programs, such as genetic counseling, where clinical teachers may only supervise 1-3 students per year. As a result, not only are SETs more confounded due to small sample sizes, but a direct conflict exists between respecting learner anonymity and providing timely and relevant feedback to faculty. In such contexts, even using SETs to evaluate the nature of the learning environment may be unreliable due to student concerns about identifiability and fear of retaliation for unfavorable evaluation. This paper will review the literature regarding SETs, barriers to this process within the clinical setting, and the unintended downstream consequences. Options for addressing issues related to the use of SETs will be considered, with particular focus on the process of reflection and the use of teaching consultations or peer support groups as a means to improve teaching effectiveness in this learning environment.

Map retrieval intention recognition based on relevance feedback and geographic semantic guidance: For better understanding user retrieval demands

Effective retrieval is essential for finding resources in demand handily amidst extensive data records in data warehouse. Mainstream map retrieval methods suffer from intention gap problem and are incapable to describe sophisticated user demands precisely due to the limits of low- and middle-level text or visual feature matching, resulting in unsatisfactory retrieval results. Such limitations are more marked when map retrieval demands were characterized with joint constraints of geographic concepts. To address this issue, we propose a map retrieval intention recognition method to perceive user demands with relevance feedback samples and geographic semantics guidance. Specifically, we construct a hierarchical intention expression model to describe retrieval goals and their multi-dimensional attribute constrains; incorporate geographic ontologies to provide semantic guidance and facilitate recognition; utilize the frequent itemset mining (FIM) algorithm Apriori to generate intention candidates from relevance feedback samples, and search for the optimal intention set by adopting the minimum description length (MDL) principle. The experiments verify the effectiveness of Apriori algorithm and MDL principle on intention recognition. The proposed method outperforms the FIM algorithm Gene Ontology (RuleGO) and the Decision Tree algorithm with Hierarchical Features (DTHF) with higher recognition accuracy and noise tolerance. Furthermore, through our sample augmentation strategy, the method yields promising recognition accuracy even when the feedback sample size is as low as ten, substantially reducing the feedback burden in human-computer interactions. We envision that the application of our method in spatial data infrastructures (SDIs), such as geoportals and catalogue services, could enhance the quality of service and user experience in geospatial data discovery.

Content-based image retrieval of Indian traditional textile motifs using deep feature fusion

In the fast-paced fashion world, unique designs are like early birds, grabbing attention as online shopping surges. Fabric texture plays an immense role in selecting the perfect design. Indian Traditional textile motifs are pivotal, showing rich cultural origins and attracting worldwide art fanatics. Yet, technology-driven abstract forms are posing a challenge for them. The decline of handmade artistic ability due to computerization is concerning. Crafting new designs associated with the latest trends is time- consuming and requires diligence. In this work an interactive CBIR (content-based image retrieval) system is presented. It utilizes deep features from InceptionV3 and InceptionResNetV2 models to match query designs with a database of traditional Indian textiles. Its performance is tested with Caltech-101, Corel-1K state-of-the-art datasets, and Indian Textiles datasets and the results are shown to be finer than the existing approaches. The similarity-based fine-grained saliency maps (SBFGSM) approach is employed to visualize the importance of features. Our approach combines deep feature fusion with PCA dimensionality reduction and speeds up search using a clustering approach. Relevance feedback is employed to refine the retrievals. This tool is expected to benefit designers by accelerating the design cycles by bridging the gap between human creativity and A.I. assistance.

Interactive AI Virtual Teaching Resource Intelligent Recommendation Algorithm Based on Similarity Measurement on the Internet of Things Platform

Abstract To tackle the issues of low recommender system correctness and the long time of recommendation in the traditional intelligent recommendation algorithm of interactive artificial intelligence (AI) virtual teaching resources, a similarity measurement-based intelligent recommendation algorithm of interactive AI virtual teaching resources is proposed. According to the contact frequency of adjacent learners, the behavior patterns of learners are mined, and the user characteristics are selected on the basis of the mutual information feature selection method. In the context of the category attribute matrix and learner category attribute scoring matrix of interactive AI virtual teaching resources, the user interest matrix is constructed by the Relevance Feedback based on Inverse Learning Function (RF-ILF) method. The user similarity is calculated and the neighbor set is found in the similar user clusters after clustering. According to the results of the simulation, the suggested algorithm is more efficient and provides more accurate recommendations when recommending interactive AI virtual teaching resources.

Document retrieval using clustering-based Aquila hash-Q optimization with query expansion based on pseudo relevance feedback

A document retrieval system helps users to retrieve the relevant documents corresponding to their query quickly and easily. In the real world, document retrieval is a difficult task due to high volumes of data, unstructured data, and different formats of data. Even though many research techniques are introduced, major problems like vocabulary mismatch and non-linear matching still need to be solved. In this work, the Aquila hash-q optimizer is the proposed matching technique with the clustering technique to retrieve the document in a time-efficient manner for the user query without collision. First, preprocessing is done by eliminating the stop words from the document, stemming, and grouping documents in a cluster into a single document using Hierarchical Density-based Sampling Spatial Cluster of Applications with Noise (HDBSSCAN) clustering. This clustering algorithm is powerful, robust to noise, and scalable and identifies clusters of documents that are related to each other. Additionally, the sampling technique used in this clustering algorithm increases the clustering speed by reducing the size of the document which improves the performance of document retrieval systems. Secondly, the queries are searched using the Aquila hash-q optimizer matching technique by which the relevant documents are retrieved. The Aquila hash-q optimization works by pre-computing a hash table of the terms in a document collection and then using this hash table to quickly identify the relevant documents from the given query. This can significantly improve the speed of document retrieval, especially for large document collections. Aquila hash-q optimization can improve the accuracy, efficiency, and scalability of document retrieval systems. The effectiveness of the Hierarchical Density-Based Clustering Aquila Optimization approach is determined by various analyses through NPL, LISA, and CACM data in terms of precision @ 5 (0.497), precision @ 10 (0.425), Mean Average Precision (MAP) (0.462) by comparing our approach with various methods. As a result, the Aquila hash-q optimizer is the proposed matching technique to retrieve the document in a time-efficient manner for the user query without collision.

Cognitive and physiological evaluation of virtual reality training in nursing

Virtual reality (VR) has increasingly been used in several disciplines, including nursing, a profession in high demand that is now facing a shortage. This study investigated the effectiveness and efficacy of VR hybrid training over traditional training (TR) methods. Sixteen college students were recruited and randomly assigned to get 100% TR or 50% TR and 50% VR (VR50). Participants attended a three-day program with a registered nurse, consisting of lectures and practical lab sessions. Participants’ performance, training time, cognitive development, physical development, mental workload, user experience, Students’ Satisfaction and Self-Confidence, and team learning were evaluated. The results showed that the VR50 performed as well as, and sometimes even better than the TR group (p-value = 0.043). VR50 group significantly had higher cognitive development and found VR easy to use and attractive (p-value < 0.05). VR-integrated training makes nurses’ training more affordable and accessible while providing instant and relevant feedback. Practitioner Summary: This study assessed employment-integrated virtual training in nursing, particularly peri-care, by comparing the performance, cognitive, physical, and mental workload of traditional and integrated VR training groups. The findings of this study provide significant support for incorporating VR training into educational settings.

275 Cultural Adaptation and Implementation of Cognitive Behavior Therapy in a Rural Mental Health Clinic of Puerto Rico

OBJECTIVES/GOALS: Culturally Centered Cognitive-Behavioral Therapy (CCCBT) is a therapeutic approach that acknowledges the effect of cultural values, beliefs, and norms on mental health. Study aims to develop a CCCBT for rural community by enhancing engagement, relevance, and outcomes through culturally tailored interventions, and evaluate its implementation. METHODS/STUDY POPULATION: Diverse mixed methods study with participants ages 18-65 from Partial Hospitalization Program in rural community located in Puerto Rico Mountain region. CCCBT intervention will have a span of 9 consecutive bi-weekly sessions. The study consists of 2 phases: 1. Evaluate intervention acceptability through focus groups (n=12) with repeated measures to learn about enhancing mental health outcomes, 2. Measure intervention effectiveness in reducing mild-moderate depression and anxiety symptoms using validated questionnaires (PHQ-9 and GAD-7) with a 100 participants from same region. Descriptive analysis for sociodemographic characteristics will be used to provide overview of participants characteristics. Qualitative analysis of focus groups will be obtained through feedback and content analysis. RESULTS/ANTICIPATED RESULTS: We anticipate high intervention acceptability through positive participant feedback and cultural relevance feedback from focal groups. We expect to find significant reduction in symptoms of depression and anxiety post-intervention, demonstrating the efficacy of CCCBT in Puerto Rican rural communities. These results will support the feasibility of implementing this approach, addressing mental health disparities, and enhancing treatment engagement and outcomes in underserved populations. The study will inform future research to refine and expand this intervention in rural communities of Hispanics and/or other rural underserved populations in the US. DISCUSSION/SIGNIFICANCE: This study’s success will advance culturally tailored mental health care in underserved rural Puerto Rican communities, reducing disparities and enhancing treatment relevance and engagement, with potential implications for improving mental health outcomes in similar populations in the US and worldwide.

Inaugural Review Prize 2023: The exercise hyperpnoea dilemma: A 21st-century perspective.

During mild or moderate exercise, alveolar ventilation increases in direct proportion to metabolic rate, regulating arterial CO2 pressure near resting levels. Mechanisms giving rise to the hyperpnoea of exercise are unsettled despite over a century of investigation. In the past three decades, neuroscience has advanced tremendously, raising optimism that the 'exercise hyperpnoea dilemma' can finally be solved. In this review, new perspectives are offered in the hope of stimulating original ideas based on modern neuroscience methods and current understanding. We first describe the ventilatory control system and the challenge exercise places upon blood-gas regulation. We highlight relevant system properties, including feedforward, feedback and adaptive (i.e., plasticity) control of breathing. We then elaborate a seldom explored hypothesis that the exercise ventilatory response continuously adapts (learns and relearns) throughout life and ponder if the memory 'engram' encoding the feedforward exercise ventilatory stimulus could reside within the cerebellum. Our hypotheses are based on accumulating evidence supporting the cerebellum's role in motor learning and the numerous direct and indirect projections from deep cerebellar nuclei to brainstem respiratory neurons. We propose that cerebellar learning may be obligatory for the accurate and adjustable exercise hyperpnoea capable of tracking changes in life conditions/experiences, and that learning arises from specific cerebellar microcircuits that can be interrogated using powerful techniques such as optogenetics and chemogenetics. Although this review is speculative, we consider it essential to reframe our perspective if we are to solve the till-now intractable exercise hyperpnoea dilemma.

Gaze-enabled activity recognition for augmented reality feedback

Head-mounted Augmented Reality (AR) displays overlay digital information on physical objects. Through eye tracking, they provide insights into user attention, intentions, and activities, and allow novel interaction methods based on this information. However, in physical environments, the implications of using gaze-enabled AR for human activity recognition have not been explored in detail. In an experimental study with the Microsoft HoloLens 2, we collected gaze data from 20 users while they performed three activities: Reading a text, Inspecting a device, and Searching for an object. We trained machine learning models (SVM, Random Forest, Extremely Randomized Trees) with extracted features and achieved up to 89.6% activity-recognition accuracy. Based on the recognized activity, our system—GEAR—then provides users with relevant AR feedback. Due to the sensitivity of the personal (gaze) data GEAR collects, the system further incorporates a novel solution based on the Solid specification for giving users fine-grained control over the sharing of their data. The provided code and anonymized datasets may be used to reproduce and extend our findings, and as teaching material.

Economics as a common interdisciplinary platform–a cybernetic approach

Cybernetics is the study of communication and control processes in biological, mechanical, and electronic systems and as such more relevant to the study of human behavior than simple paradigms, for instance Bounded Rationality (BR). Cybernetic Economics has like other cybernetic systems three levels: Economics aims to maximize the utility of goods and services Relevant Feedback is QALY-effects, for instance: 1) Eco effects, 2) Human relations and 3) Personal income System Management: Ad 1: A carbon neutral economy by a global CO2-emission Tariff (ET) Ad 2: Operation of the Big5 by Neuroeconomic Model (NeM) Ad 1-2: Informal bottom-up support of tripartite management (3P) Ad 3a: Market-based economic growth Ad 3b: Equalization of the market-based income by Universal Basic Income (UBI) Conclusion: A test course indicates that Cybernetic Economics can be disseminated to laymen. The major dissemination challenge is to give specialized researchers an introduction course in Cybernetics. Dissemination of Cybernetic Economics should have top-priority, because it's critical to better democratic guidance of Mankind towards the best possible QALY

Relevance Feedback with Brain Signals

The Relevance Feedback (RF) process relies on accurate and real-time relevance estimation of feedback documents to improve retrieval performance. Since collecting explicit relevance annotations imposes an extra burden on the user, extensive studies have explored using pseudo-relevance signals and implicit feedback signals as substitutes. However, such signals are indirect indicators of relevance and suffer from complex search scenarios where user interactions are absent or biased. Recently, the advances in portable and high-precision brain-computer interface (BCI) devices have shown the possibility to monitor user’s brain activities during search process. Brain signals can directly reflect user’s psychological responses to search results and thus it can act as additional and unbiased RF signals. To explore the effectiveness of brain signals in the context of RF, we propose a novel RF framework that combines BCI-based RF with pseudo-relevance signals and implicit signals to improve the performance of document re-ranking. The experimental results on the user study dataset show that incorporating brain signals leads to significant performance improvement in our RF framework. Besides, we observe that brain signals perform particularly well in several hard search scenarios, especially when implicit signals as feedback are missing or noisy. This reveals when and how to exploit brain signals in the context of RF.

Feature importance feedback with Deep Q process in ensemble-based metaheuristic feature selection algorithms

Feature selection is an indispensable aspect of modern machine learning, especially for high-dimensional datasets where overfitting and computational inefficiencies are common concerns. Traditional methods often employ either filter, wrapper, or embedded approaches, which have limitations in terms of robustness, computational load, or capability to capture complex interactions among features. Despite the utility of metaheuristic algorithms like Particle Swarm Optimization (PSO), Firefly Algorithm (FA), and Whale Optimization (WOA) in feature selection, there still exists a gap in efficiently incorporating feature importance feedback into these processes. This paper presents a novel approach that integrates the strengths of PSO, FA, and WOA algorithms into an ensemble model and further enhances its performance by incorporating a Deep Q-Learning framework for relevance feedbacks. The Deep Q-Learning module intelligently updates feature importance based on model performance, thereby fine-tuning the selection process iteratively. Our ensemble model demonstrates substantial gains in effectiveness over traditional and individual metaheuristic approaches. Specifically, the proposed model achieved a 9.5% higher precision, an 8.5% higher accuracy, an 8.3% higher recall, a 4.9% higher AUC, and a 5.9% higher specificity across multiple software bug prediction datasets and samples. By resolving some of the key issues in existing feature selection methods and achieving superior performance metrics, this work paves the way for more robust and efficient machine learning models in various applications, from healthcare to natural language processing scenarios. This research provides an innovative framework for feature selection that promises not only superior performance but also offers a flexible architecture that can be adapted for a variety of machine learning challenges.

WHAT HINDERS INTERNET OF THINGS (IOT) ADOPTION IN THE CHINESE CONSTRUCTION INDUSTRY: A MIXED-METHOD

Although the Internet of Things (IoT) has aroused much interest as a potential approach for improving various construction activities, the extent of its adoption remains limited. The multiple barriers that prevent the wider adoption of IoT in the construction industry need detailed investigation. However, limited research has attempted to understand the barriers to IoT adoption. Therefore, this study aims to identify the critical barriers to IoT adoption in the construction industry and explore the prioritization and hierarchical structure of the barriers factors. Data were collected from relevant literature and feedback from Chinese industry experts, sixteen barriers against IoT adoption were identified and categorized based on the TOE framework assessed in the construction industry. An integrated interpretation structure model and decision-making and trial evaluation laboratory (ISM-DEMATEL) approach is adopted to analyze the interdependence between identified constructs and their intensities. In addition, the identified constructs are also clustered into a suitable group using MICMAC analysis. Results show that inadequate infrastructure, lack of governance, and top management support are the fundamental barrier against IoT adoption. By revealing the mutual relationships and interlinking of barriers, this study will help researchers and practitioners in the construction industry to focus on strategic efforts to overcome these obstacles to effective IoT implementation. This research revealed the barriers to IoT implementation in the Chinese construction industry. Also, it provides methodological tool references for exploring the impact factor of other similar innovative technology applied in this industry.

Nitrous oxide inhibition of methanogenesis represents an underappreciated greenhouse gas emission feedback.

Methane (CH4) and nitrous oxide (N2O) are major greenhouse gases that are predominantly generated by microbial activities in anoxic environments. N2O inhibition of methanogenesis has been reported, but comprehensive efforts to obtain kinetic information are lacking. Using the model methanogen Methanosarcina barkeri strain Fusaro and digester sludge-derived methanogenic enrichment cultures, we conducted growth yield and kinetic measurements and showed that micromolar concentrations of N2O suppress the growth of methanogens and CH4 production from major methanogenic substrate classes. Acetoclastic methanogenesis, estimated to account for two-thirds of the annual 1 billion metric tons of biogenic CH4, was most sensitive to N2O, with inhibitory constants (KI) in the range of 18-25μM, followed by hydrogenotrophic (KI, 60-90μM) and methylotrophic (KI, 110-130μM) methanogenesis. Dissolved N2O concentrations exceeding these KI values are not uncommon in managed (i.e. fertilized soils and wastewater treatment plants) and unmanaged ecosystems. Future greenhouse gas emissions remain uncertain, particularly from critical zone environments (e.g. thawing permafrost) with large amounts of stored nitrogenous and carbonaceous materials that are experiencing unprecedented warming. Incorporating relevant feedback effects, such as the significant N2O inhibition on methanogenesis, can refine climate models and improve predictive capabilities.

Interactive semi-supervised learning for microarray analysis

Microarray technology has generated vast amounts of gene expression data with distinct patterns. Based on the premise that genes of correlated functions tend to exhibit similar expression patterns, various machine learning methods have been applied to capture these specific patterns in microarray data. However, the discrepancy between the rich expression profiles and the limited knowledge of gene functions has been a major hurdle to the understanding of cellular networks. To bridge this gap so as to properly comprehend and interpret expression data, we introduce Relevance Feedback to microarray analysis and propose an interactive learning framework to incorporate the expert knowledge into the decision module. In order to find a good learning method and solve two intrinsic problems in microarray data: high dimensionality and small sample size, we also propose a semi-supervised learning algorithm: Kernel Discriminant-EM (KDEM). This algorithm efficiently utilizes a large set of unlabeled data to compensate for the insufficiency of a small set of labeled data and it extends linear algorithm in Discriminant-EM (DEM) to kernel algorithm to handle nonlinearly separable data in a lower dimensional space. Relevance Feedback technique and KDEM together construct an efficient and effective interactive semi-supervised learning framework for microarray analysis. Extensive experiments on the yeast cell cycle regulation data set and Plasmodium falciparum red blood cell cycle data set show the promise of this approach.