Measuring similarity between two text snippets is one of the essential tasks in many NLP problems and it has been still one of the most challenging tasks in the field. Various methods have been proposed to measure text similarity. This survey reviews more than 150 of the related papers, introduces a comprehensive taxonomy with three main categories, and discusses the advantages and disadvantages of these methods. The first category is lexical methods that only focus on text pair’s surface similarity. These methods consider the text as a sequence of characters, tokens, or a mixture of these two. Some recent studies use deep learning techniques for detecting lexical similarity in alias detection task. The second category is semantic methods that take into consideration the meaning of the words based on some pre-prepared knowledge-bases like Wordnet or using Corpus-based methods. Some recent studies use modern deep learning techniques like transformers and Siamese networks to create document embedding that outperform other methods. The final category is hybrid methods that take advantage of all other methods even syntactic parsing in some cases. Note that high-quality syntactic parsers are not present for many languages and that using them has some side-effects on performance and speed.

Online social networks like Instagram are places for communication. Also, these media produce rich metadata which are useful for further analysis in many  elds including health and cognitive science. Many researchers are using these metadata like hashtags, images, etc. to detect patterns of user activities. However, there are several serious ambiguities like how much reliable are these information. In this paper, we attempt to answer two main questions. Firstly, are image hashtags directly related to image concepts? Can image concepts being predicted using machine learning models? The results of our analysis based on 105000 images on Instagram show that user hashtags are hardly related to image concepts (only 10%of test cases). Second contribution of this paper is showing the suggested pre-trained model predicate image concepts much better (more than 50% of test cases) than user hashtags. Therefore, it is strongly recommended to social media researchers not to rely only on the user hashtags as a label of images or as a signal of information for their study. Alternatively, they can use machine learning methods line deep convolutional neural network model to describe images to extract more related contents. As a proof of concept, some results on food images are studied. We use few similarity measurements to compare result of human and deep convolutional neural network. These analysis is important because food is an important society health  eld.

A spatial trajectory is a record of moving object’s spatial changes through time and is modeled by a sequence of discrete points with spatio-temporal coordinates. Increasing number of moving objects and positioning technologies resulted in immense number of spatio-temporal data needing various analyses. Extracting similar trajectories is one of the crucial analyses in spatial trajectories. So far various distance functions have been proposed for measuring similarity where each one has addressed similarity from its own point of view and is suitable for particular data with special characteristics. Thus, functions effectiveness is not the same for all kind of data and applications and understanding capabilities and characteristics of functions is the prerequisite of choosing the suitable function. In this paper, a comparative experimental study is conducted on the effectiveness of seven widely used trajectory similarity measures which are the base of many other former proposed distance functions and their advantages and drawbacks are discussed.

Movement of objects is taking place in geographical contexts. Context directly/indirectly influences movement process and causes different reactions to moving objects. Therefore, considering context in movement studies and the development of movement models are of vital importance. In this regard, incorporating context can play a crucial role in similarity measurement of objects movements and their corresponding trajectories. Trajectories of moving point objects, beside their spatial and temporal dimensions, have another aspect which is called contextual dimension. This dimension, however, has been less considered so far and a few researches in trajectory analysis domain have investigated it. To this end, this research develops a method based on Euclidean distance in which individual spatial, temporal, and contextual dimensions as well as their integration can be explored in the process of similarity measurement of trajectory. Beside the simplicity of the method, it is developed in a way for taking into account every small change in each type of dimension (s). To validate the proposed method and survey the role of contextual data in similarity measurement of trajectories, three experiments are performed on commercial airplane dataset. Accordingly, geographical coordinates and altitude of airplane as spatial dimension, travel time as temporal dimension, and airplane speed, wind speed, and wind direction as contextual dimension are utilized in these experiments.The first experiment measures the correspondence of trajectories in different dimensions. Also, it explores the role of dimensions weights individually and collaboratively along the similarity measure process. The results demonstrate that weights severely affect similarity values, while they are totally application dependent. Meanwhile, it can be confirmed that contexts may increase or decrease the values of trajectories similarities. This effect can be seen in the average of relative similarity values of commercial airplanes trajectories in spatial (0.60), spatial-temporal (0.51), and spatial-temporal-contextual (0.46) dimensions. Contexts can enhance and restrict movements as well. To justify this statement, the second experiment is conducted to explore how movement and geographical contexts interact in similarity measure process. To this end, four sample trajectories are compared with respect to different dimensions. For a pair of trajectory, the relative similarity value at spatial dimension is 0.04. By incorporating time dimension, this value increases to 0.30 at spatio-temporal dimension. Given the high similarity of these two trajectories in wind direction, wind speed, and airplane speed (0.85), the ultimate similarity of them becomes 0.48. In contrast, for another pair of trajectory, the spatial and spatio-temporal similarity values are 0.85 and 0.91, respectively. Considering the similarity value of these two trajectories in wind direction, wind speed, and airplane speed (0.37), the final relative similarity becomes 0.73. The third experiment sought for the role of motivation context in similarity measure process. Although such context is very difficult to capture and in many applications will remain inaccessible, we consider the pilots decisions in handling the airplanes during the approaching and landing phases (i.e., continuous descent final approach or dive and drive) as the motivation context in this application. Choosing either of these techniques highly affects the figure of trajectories where quantifying them can be accomplished by measuring the similarity of trajectories at spatial and spatial-temporal dimensions. All in all, the results of the above experiments yield the robustness of the proposed method in similarity measurement of trajectories as well as its sensitivity to slight alterations in dimensions.

The electric arc is one of the most intense electrical events. This phenomenon occurs due to the electric discharge between two conductors or between a conductor and the ground, through the air. When the short-circuit current intensity is high, it can be easily detected by traditional protection equipment. However, when the short-circuit current is low, traditional protection methods cannot detect these faults. Faults that do not generate enough fault current to be detected by conventional protective equipment are called high-impedance faults (HIFs). HIFs can cause serious safety hazards in power distribution systems and damage to equipment due to the risk of arc ignition. This paper presents a new detection scheme for HIFs in electrical distribution systems based on similarity measurement. In this method, based on the waveform of two consecutive half-cycles of the current, an index is extracted that can be used to detect HIFs. The proposed HIF detection algorithm can distinguish these events from other non-fault events with waveforms that may be similar to HIF waveforms. In this paper, four case studies are simulated to verify the proposed HIF detection algorithm. The simulation results demonstrate the acceptable performance of the proposed method in detecting HIFs and distinguishing them from other events.

Tropical Cyclones are a natural and complex phenomenon that threatens the life and property of human society. The accuracy of predicting their trajectories is critical to reducing economic loss and saving human lives. When a storm occurs, context information such as wind speed and intensity, air pressure, storm direction, water surface temperature, etc. are effective in changing the direction of storm trajectory. Accordingly, considering this informations can improve the forecasting accuracy. Researchers have used various methods to predict the direction of hurricane movements to achieve the highest accuracy in forecasting. Recently, deep learning methods have shown a potential capability to process complex data efficiently and accurately. In this paper, we used the Long Short-Term Memory method to predict the future path and location of tropical cyclones in the North Atlantic Ocean by measuring the similarity of tropical cyclone trajectories and taking into account positional parameters and context information such as wind speed and storm direction. The obtained results show an improvement in the accuracy of the prediction compared to the lack of context information for the 3, 6, 9, and 12 hours time periods. The distance between the predicted trajectory path and actual trajectory path has been reduced from 1. 9 to 4. 5 km, taking into account the context information.

Gall forming is one of the various damage forms of insects and mites on plants. This study carried out in collecting and identification of willow pests and their natural enemy’s fauna in West Azerbaijan province in a research project during 2010-2014. The fauna of willow gall makers were evaluated at 15 day intervals starting in May to October. In order to sampling, 4 branches of each tree were studied in four geographical directions. All statistical procedures were performed using the SPSS 19 software. Susceptibility and resistance of willow species to two important gall pests were determined in Saatloo station. A total of 10 species of gall inducing pests on willow trees were collected. The results showed that Salix excelsa S. G. Gmel and Salix triandra Linnaeus were susceptible species to Pontania vesicator Bremi-Wolf and Rabdophaga heterobia Loew., respectively. The highest similarity indexes (Sorensen and Jacard) of gall inducing factors of willow trees were recorded between Salmas and Darre Khan (Urmia - Ghasemloo).