Research and Projects

Modelling of Location Based Social Networks Based on the Internet of Things

Nowadays, social networks (SNs) can be considered an important part of human life. Since the advent of SNs, different types of SNs such as multimedia SN, Event-Based SN, and Location-Based SN have been applied extensively by people every day. Although users and their social interactions is the main focus in SNs, additional dimensions such as location and contexts have been developed to provide a wide variety of customized and personalized services. Nowadays, with tremendous growth of the Internet of Things (IoT), smart devices are becoming more intelligent so that they can provide. For example, using data provided by smart speaker, e.g., Google Home, user preferences in different areas such as movies, music, and events can be algorithmically extracted. With the potential of IoT devices, user profiles including users’ behaviors in a wide variety of aspects can be available. In other words, a deeper understanding of user behaviors in addition to their social behaviors can now be available using IoT devices without human interventions. In this project the ultimate goal is the use of embedded resources in mobile devices, especially smart phones, to improve location based services and course queering systems. Due to the use of direction and azimuth information, this new services are called location and orientation aware services. Mobile Geo Information System (MGIS) users by utilizing of various ways can interact with physical objects. One of these ways is pointing. In this study, by similar manner we want to use mobile devices as a pointer for obtaining information from buildings and land use of floors of the buildings.

1. 
   
2. 
   
3. 
   
4. 
   
5. 
   
6. 
   

One of the most critical issues in urban management is incidence management. Most incidences unexpectedly occur which makes it difficult to be predicted. Consequently, designing a real-time alarming system to support first responders and people is vital. As the most important part of alarming systems is information, most of the existing solutions for alarming incidences are based on either Internet of Things (IoT) or User Generated Spatial Content (UGSC). However, IoT suffers from inadequate data distribution and UGSC does not provide high-quality data. The main objective of this project was to design and develop an architecture to integrate these two information network models to provide more accurate alarms for incidences. To do so, the issues of the IoT and UGSC integration were addressed and an integrated cloud computing architecture was proposed to resolve them. The main core of the architecture is based on a pub/sub-event driven computing model which is integrated with data-driven external computing nodes. The main contributions of this thesis are to deliver an integrated event-driven/data-driven architecture that delivers more accurate alarms, propose a novel approach to filter high-quality UGSC and IoT data, and reduce the energy consumption by Integrated Power Aware Wireless sensor network and Location-Based Social Network (IPAWL). In general, the proposed architecture is highly recommended to be used for detecting complex incidents in large-scale projects.

1. 
   
2. 
   
3. 
   
4. 
   

Governments face many challenges today that address rough problems in solving some of them. Basically, some of the challenges have features that require a lot of resources to solve. Solving some of the natural and social crises and issues arising from them are considered by this group. Crises are usually large, and the control of their implications requires the use of a large amount of human and financial resources. Various solutions are available to control these crises. Using power of people who do not require much expertise or special sensitivity is one of the possible options. Despite some negative points about the crowd of people such as low-level responsiveness or expertise, such remarkable privileges as high numbers, wide spatial coverage and high diversity of capabilities make them a special force to solve some of the crises and other applications. Therefore, it is possible to create a lot of forces with the unity of a few people’s capabilities, and adapt it to the conditions. Yet the important thing is how to manage and unite these forces to resolve crises at the level of a vast area. The use of masses of people is largely seen in discussions on product quality and design ideas or any specific products or even special applications such as urban planning. In these topics, the intellectual potential of individuals in an e-commerce space is shared. It is important that in such a space, there is no physical practice and no time urgency. However, the conditions required to solve problems in the crisis are different. In a crisis area, many problems have been encountered in wide areas that should first be identified quickly. Popular forces with different capabilities in the region should be identified and lead to relief priorities based on their position and capabilities. They should also be informed of the outcome of their actions, and if one of these people could not solve the problem, alternate or complementary persons should immediately introduced. In this regard, a kind of information service is introduced in a crowdsourcing form. The service operates in the context of a compound system and, with the help of elements of this system, can quickly identify the crisis and identify the crowd of people in accordance with their expertise, capabilities and the standard crisis management guidelines, in order to solve the problems of the region.

1. 
   
2. 
   
3. 
   
4. 
   

Natural disasters have always been one of the threats to human societies. Earthquake is one of the most unpredictable events that could lead to a great human catastrophe. The damage map can determine the priority of the relief workers. It can also be more efficient in resource allocation. One of the main resources in mapping the damage is remote sensing. Methods based on remote sensing usually have a time delay of 48 to 72 hours. Remote sensing data has spatial, temporal and spectral constraints. In recent years, due to the potential of user-generated spatial content, they have been used more often in crisis management studies. Although the detection of earthquakes via of user-generated spatial content in a short time can provide effective results, the use of tools such as seismographs or accelerators allows for more accurate results. Considering that earthquake’s sensor networks can not quantify damages, it can be solved by integrating social networking information and other information sources. So, when combining the results of user generated data with physical sensors, it is possible to use the benefits of both. The main objective of this research is the earthquake damage estimation, based on the integration of different data sources with the user's spatial content. In this study proposed an approach that applied social media data for the earthquake damage assessment at the county, city, and 10 x 10 km grids spatial levels using Naïve Bayes, Support Vector Machine (SVM), and Deep Learning classification algorithms. In this study, classification is evaluated using overall accuracy, precision, recall, and F-score metrics. Then for understanding the message propagation behavior in the study area, temporal analysis based on classified messages will be done. Also, variability of spatial topic concentration in three classification algorithms after the earthquake will be examined using Location Quotation (LQ). In addition, the damage map based on the result of the classification of the three algorithms into three spatial levels will be created. For validation, confusion matrix metrics, Spearman’s rho, Pearson correlation, and Kendall's tau will be used. In this study, binary classification and multi-class classification have been done. Binary classification is used to classify messages into two classes of damage and non-damage so that their results can finally be used to estimate the earthquake damage. Multi-class classification is used to categorize messages to increase post-crisis situational awareness. In binary classification, the SVM algorithm performed better in all the indices that gain 89.30% overall accuracy, 81.22 F-measure, 79.08% recall, 85.62% precision, and 0.634 Kappa. In a multi-class classification SVM algorithm performed better in all the indices that gains 90.25% overall accuracy, 81.22% F-measure, 88.5% recall, 93.26% precision, and 0.825 Kappa. Based on the results of the temporal analysis, most of the damage related message was related to the earthquake day and decreased in the following days. Also, most of the messages relates to infrastructure damages and injured, dead and missing people, are reported on the day of the earthquake. In addition, results of LQ indicated Napa as an epicenter of the earthquake is considered as the concentration of damage-related messages in all algorithms. This indicates that our approach has been able to identify the damage well and has considered the earthquake epicenter one of the most affected counties. The findings of the damage estimation showed that going away from the epicenter lowered the amount of damage. Based on the result of the validation of the estimated damage map with official data, the SVM performed better for damage estimation followed by Deep Learning. In addition, at the county spatial level, algorithms have better performance with Spearman’s rho of 0.8205, Pearson correlation of 0.5217, and Kendall's tau of 0.6666. In the case of seismic data, a series of seismic data related to the Napa earthquake provided by the United States Geological Survey is used to estimate the amount of damage caused by the earthquake. This information contains the recorded magnitude of the earthquake and the amount of damage caused by it in different locations caused by the Napa earthquake. In this research Support Vector Machine (SVM) and multivariate nonlinear regression method were used for damage estimation using seismic data. Multivariate nonlinear regression method performed better in terms of correlation coefficient (0.75). In this research, a knowledge-based method based on pixel level information was used to evaluate the ability of the ANFIS model to determine the amount of destruction of buildings in the earthquake area. So in this research combination of social network data, seismic network and remote sensing was used to assess the earthquake damage. Due to that no damage could be detected using remote sensing images in the Napa earthquake; the combination of social network and seismic data was finally used. The results of the integrated damage model at the city spatial level showed that 4 out of 5 cities with the highest damage were the same between the integrated model of damage estimation and official data. Based on the result of the validation of the integrated estimated damage map at the city spatial level with official data, Spearman's rho (from 0.41 to 0.8), Pearson correlation (from 0.53 to 0.76) and Kendall's tau (from 0.36 to 0.66) indices have increased relative to damage estimation based solely on social network data. The results of the integrated damage model at the county spatial level showed that 3 out of 5 counties with the highest damage were the same between the integrated model of damage estimation and official data. Based on the result of the validation of the integrated estimated damage map with official data, Spearman's rho, Pearson correlation and Kendall's tau indices were almost the same relative to damage estimation based solely on social network data.The results of integrated damage estimated map from social network and seismic data at the 10 x 10 km grids spatial level showed that with increasing distance from the epicenter of the earthquake, the intensity of earthquake damages is reduced. Based on the result of the validation of the integrated estimated damage map at the 10 x 10 km grids spatial level with official data, Pearson correlation (from 0.16 to 0.63) and Kendall's tau (from 0.82 to 1) indices have increased and Spearman's rho (0.92 for social network and 0.86 for integration) was almost the same relative to damage estimation based solely on social network data. Therefore, based on the results of this study, it can be concluded that the combination of different data sources can improve performance in damage estimation.

1. 
   
2. 
   
3. 
   
4. 
   
5. 
   

Predicting human trajectory in different spaces is critical.trajectory prediction means predicting locations in which a person will be in the future. For instance, a correct trajectory prediction for guiding robots and other autonomous vehicles is significant; since an incorrect trajectory prediction may lead to severe collisionsand endanger its safety. The environment can be divided into outdoor and indoor categories. In indoor environments, people have less space to move and trajectories, in addition to being closer, are affected by the spatial characteristics of the building. The aim of this study is to predict human trajectory in the indoor of a building using deep learning methods. The method used is based on data-driven methods of recurrent deep learning networks and in particular, long short-term memory networks (LSTM). In the proposed network, factors including past trajectory, fixed and moving obstacles, neighbors, spatial characteristics and time are considered. Each of these factors has a special representation. In order to improve the accuracy of trajectory prediction in this research, data augmentation and transfer learning methods have been used. According to the results of this study, the proposed deep learning network reduces the prediction error of the trajectory in indoor spaces with an average distance error of 18 cm for 12 prediction points. In crowded environments, this network's accuracy was tested, and the average distance error at 12 prediction points was 31 cm.This network outputs trajectories without collision with obstacles and neighbors. Furthermore, it considers the spatial characteristics of the environment and predicts possible changes in direction when a person encounters a wall.

1. 
   
2. 
   
3. 
   

Wheat is a strategic crop and an essential source of food. It also plays a major role in the market and business of millions of farmers worldwide. Given the importance of costs in wheat production, its damage factors should be identified and managed. The Sunn pest is among the most prominent damage factors of wheat in Iran. The damage caused by this pest reduces the nutritional value of wheat and destroys a great deal of this crop. In recent decades, the integrated pest management (IPM) has been used as a comprehensive strategy in agriculture. Based on a natural approach, the IPM benefits from different managerial tools to reduce the use of chemicals for pest control through various solutions, a major stage of which is continuous monitoring. This study aims to design and implement a volunteered geographic information system to control the Sunn pest. In the proposed method, users provide a spatial database with all or part of the information about the Sunn pest spots including images, coordinates, altitudes, and names of regions or farms. Location accuracy, descriptive accuracy, and integrity were employed to evaluate the spatial data. The descriptive accuracy, location accuracy, and data integrity were then reported 77.6%, 8.6 m, and 77%, respectively, on average. Despite altitudinal differences between the higher points reported by volunteers and the reference points, the altitudinal analysis indicated that the reported points were efficient in evaluating the outbreak of the Sunn pest. The altitudinal analysis also revealed that the volunteers recorded and presented less information regarding the overwintering sites than the reference points.

1. 
   
2. 
   

With the advances of communication networks, innovative equipment, and their omnipresence, demand for generating more sophisticated location-based systems such as wayfinding, routing, and positioning services, has been increasing regularly. But these services face some limitations in places where the user is not familiar with his or her surroundings. Meanwhile, Location-based social networks (LBSN) have become a valuable resource for researchers through having much information about their users. By analyzing users’ data, their interests could be learned and various services provided in a personalized manner. In this paper, we present a method that extracts the Multiple Point of Interests (MPOI) of the user from the LBSN based on the user contexts and contextually improves the navigation service. MPOIs of the users can help them to get an understanding of their position. We provide a novel context-aware personalized wayfinding service with the help of user MPOI’s as well as landmarks alongside the road. We are going to navigate the user along the road in a contextualized manner. One of the essential uses of this method is Reverse Geocoding, or the definition of the address according to the user's position. To adapt location-based services to user interests, the selection of POIs depends on the user's environmental, physical, spatial, and temporal conditions. This condition is called context, and the service would be context-aware.Check-ins from LBSNs have been utilized to improve location-based services (LBS), such as wayfinding. Three factors, namely personal interest, social interest, and time, are selected based on check-ins. A machine learning algorithm is used to predict user interests in all categories and context-aware mapping to dynamically display the map to the users. A dataset of Foursquare check-ins across two US cities, i.e., Los Angeles and New York, are used to realize the idea. The results demonstrate the high accuracy of the model in comparison with other practical factors of predicting interests.

1. 
   
2. 
   
3. 
   

Using 3D spatial data, causes efficiency in urban planning and faster decision making. Within the last years, such terms like Volunteered Geographic Information (VGI) and User Generated Geographic Content (UGGC) appeared, and produce a completely new phenomenon in public collecting spatial data and define a new source of data. Adding 3D data to VGI is an important step. It seems, there is not an interoperable method for collecting and sharing 3D data in the current VGI. The purpose of this study is to provide guidelines in VGI systems for preparing 3D building models. Therefore in this study, "Wall", "Roof", "Door" and "Window" were considered as the fundamental building structural elements. Then we have developed a hierarchical classification based on ontology, for describing the interaction of these elements. A methodology for collaborative development 3D building models is suggested. Within the framework of our work, users are able to add 3D information in VGI interactively. On the other hand, is looking for ways to increase non-expert users’ participation. For this purpose, a user can upload pictures of buildings that are used for texture, then as well as digitizing aerial images, digital images are superimposed on the buildings. It causes that data entry to be adequate. Also In this thesis, issues related to the preparation a digital elevation model has been explained and the use of VGI for preparing and updating DEM has been made. The lack of public participation to gather 3D information about terrain is evident. To implement our method, at first we develop a website to collect elevation data recorded by GPS users in an small area of Tehran. SO by using EGM2008 Geoid model, obtain orthometric height. To evaluate the user generated Elevation data, we compared this with SRTM and NCC elevation data set. The results showed that, VGI elevation data RMSE is more than SRTM RMSE. On the other hand both of SRTM and VGI elevation data in dense trees areas have large errors. The accuracy of VGI elevation data in streets with less building density is better than SRTM data. It seems if GPS signals are treated to disruptions caused by construction barriers and trees, can be more accurate than SRTM data. Therefor we can use combination of VGI elevation data with SRTM to create an improved free DEM in Volunteered geographic information environments.

1. 
   
2. 
   
3. 
   
4. 
   
5. 
   
6. 
   
7. 
   

The landscape is an array of related features in a specific area. Landscape description is a term given to landscape science in ecological, visual and acoustic matters. Looking at the main set of attributes used in landscape descriptions, visibility is always a main factor for decision making in visual impact assessment, as well as in other applications in aesthetic analysis, visual pollution, way finding, and landscaping. This paper attempts to show how Volunteered Geoinformation System takes parts in landscape descriptions by people's active, perceptual engagement in the word. In general term, dealing with the environmental visual pollution, finding one’s way in the spatial extended area and finding the landscape appropriating to the one’s emotional status, are some objectives which require well understood of landscape properties. Despite spatial configuration and composition of spatial relations are needed to describe surrounded environment, it still remains absent from cross-cultural tendency information to obtain how they percept landscapes. From the other hand Nowadays User Generated Environments (UGE) such as Volunteered Geographic Information (VGI) environments create a proper place to elicit people’s idea and increase their information and knowledge about the problems. Then it is needed an interdisciplinary effort, involving spatial cognitive science, geographical information system (GIS), VGI system, landscape science and computer science, to develop a participatory space to collect people’s attitude about their surrounding landscape. Spatial cognition data is used to make communication between computers and human, based on human reasoning. This paper presents some metrics which support visual aspect of landscape in plain sight alongside human mental description and spatial metrics, namely time, eye-catching, distance, solid angle, directional relations, and topological relations. Then, the six selected metrics are evaluated by asking through questionnaire from participants that if these metrics are applicable or not and about the level of importance between them. The questionnaire result shows that about more than %86 of participants do agree with these metrics. Also, According to the opinion of the participants, the proposed category of these metrics could describe landscape more than %65 due to questionnaire results. Among these metrics, it is needed to simplify the solid angle meter to be used by participant correctly. Due to this matter, the meter percentage of area is offered. Result shows that, by using linear regression and with the error of 0.0681 and correlation coefficient of 0.981, the proposed meter could be used instead of solid angle. More and more implementation of location based services (LBS) to attract consumers to purchase their product or services. However, using LBS may raise privacy concern for consumer but people increasingly make use of tools which apply the concept of location. This study compares the colour distribution in every direction of specific landscape to detect visual pollution caused by inappropriate colour distribution. one algorithms of visual data mining are handled here. The output of these algorithms is used to recommend those involved to produce such volunteered data to suggest them the most appropriate place depending to their psychological behaviors, and mental and physical disease. This state encourage people to participate more, and by doing so, a comprehensive data base will be developed.

1. 
   
2. 
   
3. 
   
4. 
   
5. 
   
6. 
   

Today, one of the most fundamental needs of people is choosing residential property. Given that, people spent most of their life in these confined spaces, such as residential property. Therefore, it seems logical, that these spaces are effective for their residents. In addition to the economic and social aspects, property influences the psyche of its inhabitants. This increases the importance and accuracy required in its selection. The aim of the present study is to describe the residential property in the context of the volunteered geographic information environments. For describing a residential property, it is necessary to consider several criteria. In this study, the extracted criteria include topological relationships, geographic location, shape, colour, directional relationships, dimensions, and height. In addition to the almost complete description about residential property, these criteria are independent from each other. According to existing surveys, the necessity of using the proposed measures was estimated as follows: geographical location 84%, topological relation 76%, directional relations 64%, height 62%, shape 60%, and color 58%, respectively. Finally, according to another survey, the proposed model was presented for two groups: one who has never seen the property, and the other who has already seen property. People who had seen the property, previously, could recall the property are more than 70 percent, and for the second group who has already been able to visualize the property before visiting it, they are about more than 65 percent. In addition, the appropriateness of the proposed criteria was surveyed with the mindset of the participants. Based on the results, the following is the appropriateness and relevant of mindset of participants: more than 90% of topological relation, 86 % of qualitative direction, 78% of the geographical location, 74% of shape-space, and 60% of height. Finally, people's satisfaction and impact of the tools used in the implemented site named SAMA was compared with three most visited Iranian sites, and a foreign site named IRANSTATE

1. 
   
2. 
   
3. 
   

Providing recommendations in cold start situations is one of the most challengingproblems for collaborative filtering based recommender systems (RSs). Although user social contextinformation has largely contributed to the cold start problem, most of the RSs still suffer from thelack of initial social links for newcomers. For this study, we were address this issue used aproposed user similarity detection engine (USDE). Utilizing users’ personal smart devices enablesthe proposed USDE to automatically extract real‐world social interactions between users. Moreover,the proposed USDE uses user clustering algorithm that includes contextual information foridentifying similar users based on their profiles. The dynamically updated contextual informationfor the user profiles helps with user similarity clustering and provides more personalizedrecommendations. The proposed RS is evaluated using movie recommendations as a case study.The results show that the proposed RS can improve the accuracy and personalization level ofrecommendations as compared to two other widely applied collaborative filtering RSs. In addition,the performance of the USDE is evaluated in different scenarios. The conducted experimental resultson USDE show that the proposed USDE outperforms widely applied similarity measures in coldstart and data sparsity situations.

1. 
   
2. 
   
3. 
   

Over time, spatial data has evolved from paper maps to Web GIS through digital mapping and finally to the current generation of GIS. Although this new interface has enhanced user’s insights about spatial information, it still needs more tangible interface that can be usable to the public and enhance human interaction with the environment and spatial objects. Combining real and virtual worlds, Augmented Reality (AR) systems can make more tangible experience with real world objects. These activities relies on the dynamic working environments. Therefore it is important to consider the users environment and its changing through context-awareness. Context-awareness is any kind of information about user’s status and its estimation allows to integrate the context with the happening changes. AR is a combination of real vision with virtual content in real –time and acts as an interface to increase user insight of the real world and interaction with it. With increasing information in the AR, the usefulness and readability of information decreases and the details and their display should also be subject to certain conditions. To overcome this problem, we offer the combination of AR with context-awareness. Hence, in this study, The AR representation varies according to the user’s context. The increasing use of sensors, their hardware and software enhancements, and growth and development of communication networks have led to development of context-aware computing. Context-aware computing is rooted in processing anywhere and anytime, and aims at understanding environmental changes in computer systems so that computers can understand and respond to their environment. The small screen, low bandwidth, interaction problems, and user's quick need to obtain response from service have driven services to provide information based on users and their environments. Nowadays, in most location planning applications, it is tried to use directional data besides location data. By using this information, better interaction and information are provided to user. The activities performed depend on user's dynamic work environment so by using context-aware knowledge, the environment and user environment changes receive more attention. The output of old programs was based on specific input and did not consider context change. Materials & Methods: By increasing the amount of realistic information, simultaneous provision of all information will not only reduce the usefulness and readability of information, but also details and their presentation. In order to overcome the problem, we have used a context-aware augmented reality that displays the necessary information according to actual increase in user's context. Speciialy in the domail of spatial services, given the massive amount of the available information and simultaneous provision of data in real-time due to small display of associated equipment make visual distraction for user, so by using context-awareness, useful and proper information can be provided to the user. Having the vulnerability maps, specifically about buildings in the region of interest is a necessary requirement for any rescue and relief teams after earthquake. Due to lack of immediate access to earthquake vulnerability maps it is necessary to provide an intelligible context information system for displaying vulnerabilities of building with the help of augmented reality. In general, any type of navigation issue, information about building vulnerabilities to select safe paths or temporary accommodation is required. In order to implement such information system, three contexts, i.e. distance, direction and time for presenting information in augmented reality was studied. To this purpose, after zoning vulnerability caused by earthquake in Tehran, Donyae-Noor business complex was used as the case study and information system was considered with three contexts for given site. In order to evaluate the presented system, our context-aware augmented reality system was compared with an augmented reality system alone. Results & Discussion: Results of evaluation show that combination of augmented reality and context-aware can reveale suitable information by considering the user context, while data presentation in augmented reality is monotonic and there is no dynamism in changing information display. The results of evaluating combined augmented reality system with context-awareness compared with augmented reality alone show that context-aware augmented reality system is more active and always reveals more adopted information to the user regarding environmental dynamics. While information is uniform and does not change in augmented reality, Using the context-aware spatial information system, the user can take different decisions under different circumstances or gain extra information about other context such as time, velocity, acceleration. To better understanding augmented reality, certain graphical forms can be used. Displaying vulnerability by augmented reality can be a tool for ordinary users or specialists in the field of urban design and management, without the need for mapping and map reading knowledge in managing important centers and building in low-risk areas. The results of integrated system evaluation from various users indicate the system's performance is superior to augmented reality system alone.

1. 
   

Improving the safety of the service needs total integrated management of the airport, which due to In-dependency of management in various sectors of activity is relatively difficult. Yet, continuance of this process, can affect all operations of an airport alone. Increasing flights in the airports lead to traffic jam of service cars results in an increase in the airport's accidents. In time of crisis in the airport, because of the lack of information on the overall situation, the decision making process will become a serious problem for managers. Hence, the safety of vehicles within the directory service turn into a rising challenge. In this study, a decision support system to provide a solution for dedicated service to aircraft carriers, enabling them to provide optimal allocation. Using this solution in times of crisis enable managers and service providers to use Geographic Information System which will be assigned to aircraft. The developed model in this study, after receiving the aircraft and service cars position, will announce that which aircraft and the service path should be served first. The implementation of proposed model at Imam Khomeini International Airport results show 36 seconds reduction of flight delays related to the real time of flight. If the proposed model be applied in several flight, it will show a better performance and a greater reduction in the total time of flight delays. Applying this model leads to increase of the power of decision-making and as a result can reduce the crisis in the airports.

1. 
   
2. 
   

Effective communication between the real and virtual worlds of the oil and gas industry is critical to prevent work problems and financial losses. Augmented reality (AR) technology can provide a solution by displaying virtual content in real-time and improving user interaction with the real world. However, using AR in location-based services can lead to problems such as illegible information on limited device screens and uniformity of information despite changing user conditions. To overcome these challenges, a context-aware AR system was developed that considers the user's environment and adapts the system's behavior accordingly. This project describes the system's architecture, selected standards, effective contexts, and modeling, and presents a prototype system developed for the Gas Company of Kerman province. The system's implementation demonstrates improved readability and efficiency over traditional AR systems and serves as a new model for AR technology in the oil and gas industry.

1. 
   
2. 
   

Contact tracing applications based on Bluetooth Low Energy (BLE) have been widely implemented during the COVID-19 pandemic. However, the use of only the Received Signal Strength (RSS) feature for proximity calculation may not be adaptable to different virus variants or scalable for other potential epidemic diseases. In this project we proposed a novel framework for evaluating and classifying personal exposure risk that considers both contact features (distance and length of contact) and environment features (crowd size and the number of recently infected cases in the environment). The framework utilizes a fuzzy expert system that is adaptable to different virus variants. The proposed method was tested on two types of viruses with different close-contact features, using four membership functions and 256 fuzzy rule sets. The proposed framework classified personal exposure risks into four classes: low risk, medium risk, high risk, and too high risk. Empirical results showed that the fuzzy logic-based approach reduced the number of false positive cases and demonstrated better accuracy and precision than the current BLE-only approaches.

Digital contact tracing is an effective strategy for preventing and controlling viral illnesses, but traditional systems face challenges from computational complexity and privacy concerns. In this project, we introduce a real-time person-to-place contact tracing system that utilizes surveillance cameras, GIS, Building Information Modeling (BIM), and image processing techniques. Our proposed system requires minimal operational resources and stores the least amount of data, eliminating the need for new privacy and security laws. By leveraging the benefits of BIM and GIS, our system is context-aware and can make decisions based on various circumstances. We tested the system on simulation videos and achieved over 80% accuracy for all models used, with a processing speed of 50 frames per second using Google Colab computational resources. Our system has the potential to be a valuable tool for public health officials to quickly identify potential sources of infection and prevent the spread of infectious diseases.

In recent years, Location-Based Social Networks (LBSNs), as a form of social media, have gained immense popularity. These networks offer location-related services, allowing users to share their experiences and the visited locations with friends across different geographical areas. The vast amount of check-in data generated within LBSNs contains rich spatial and social information, providing a valuable opportunity for researchers to study users' social behavior from a spatiotemporal perspective. LBSNs, in addition to the information available in traditional social networks, also include location information for users and generated content, making them more complex. A significant challenge lies in uncovering the architecture of these complex networks and developing suitable analytical tools. Researchers are striving to analyze various networks, including social networks, using geometric analysis, such as curvature. Essentially, curvature allows us to gain deeper insights into structure, dynamics, and evolution of networks. This thesis focuses on analyzing LBSNs by leveraging the geometric concept of curvature. Local curvature is employed to understand the structure and discover communities within the LBSN, and a user-based collaborative filtering recommendation system is developed based on the discovered communities. Subsequently, global curvature is used to link prediction between users through a hyperbolic neural network, based on spatial-temporal features. According to the obtained results, local curvature in the community detection process performs, on average, 14.33% better than other community detection algorithms. Link prediction using global curvature through a hyperbolic neural network also performs 37.7% better than the traditional method without considering global curvature and the Multilayer Perceptron (MLP) neural network.