Log Dataset Research Articles

Understanding Factors Influencing Click-Through Decision in Mobile OTA Search Engine Systems

Mobile commerce has changed the decision environment for users who intend to reserve a preferred hotel. This study aims to investigate the factors that affect the dynamic click-through decision (CTD) in mobile online travel agency (OTA) search engines. We propose a dynamic Bayesian inference framework to model individual-level users’ CTDs and examine the effects of item position, price, search cost, and the use of refinement tools. The study uses real-world search log datasets from a global OTA for both mobile and desktop searches. Our results show that (1) the primacy effect is weaker and the effect of item-ranking positions is non-linear in a mobile OTA search compared to a desktop OTA search. Mobile users pay the most attention to the top-ranking results and are less likely to click through the middle or bottom results. (2) Hotel prices have a positive effect on mobile CTDs in the whole mobile searching journey. Additionally, mobile users also tend to seek out hotels with lower price rankings on the current search engine result page. (3) The search cost, measured by the cumulative time duration, has a positive impact on mobile CTDs. The use of refinement tools enhances the effect of search cost. This study extends previous research on position and price effects in an online consumer search from PC-based internet to mobile devices. It also provides managerial implications for mobile OTA search engine marketing and investment for bidding ranking positions.

Structural Styles and their Suitability for Hydrocarbon in Eastern Sindh Monocline, Lower Indus Basin, Pakistan

The present study is aimed to recognize the structural styles of hydrocarbon exploration and petrophysical properties of the LGF. The 2-D seismic and well log data set is composed of three seismic lines and well log data of Chak 66–1. Five horizons have been marked with the help of well to seismic tie namely tops of Ranikot, Parh, upper Goru, lower Goru, basal and top massive sands, and out of which basal and massive sands are the objective focus of the present study. Based on seismic data, the study area is characterized by normal faults showing NW–SE dipping trend. Horst and graben structural features are prominent on the seismic lines, which indicated the extensional tectonic regime. Time–depth contour maps and 3D surfaces of objective horizons depict their actual spatial distribution in this area. Wireline logging analysis revealed the physical properties of both basal and massive grains of sand, as 17% effective porosity, 25% average volume of shale (Vsh), 40% water and 60% hydrocarbon saturations for basal sands, Whereas, 16% effective porosity, 35% Vsh, 30% water and 70% hydrocarbon saturations are interpreted for massive sands. Cross–plots of Nphi–Dt and Dt–Rhob also identified that both areas of sand are clean, gas–saturated and have the potential to produce hydrocarbons.

Using AI tools to fill an incomplete well log dataset: A workflow

One issue commonly found when working with well log data is the irregular abundance/availability of the different recorded parameters. This is especially applicable when working with datasets collected in different campaigns that may span through the years, even decades, or different companies. Artificial Intelligence may be useful to fill gaps in the original database, resulting in a more complete, standardised one. In this work we present a workflow that can be followed to fills gaps in a dataset using different AI techniques. It consists of four main steps: 1) feature combination selection; 2) hyperparameter tuning; 3) performance assessment and best option choice; 4) blind testing. The process can be performed iteratively, successively populating the database with missing parameters, starting with those for which there are more available training data and whose results are more reliable. In this work, we present an example in which we filled an incomplete dataset consisting of wells provided by the UK National Data Repository (NDR) of the Oil & Gas Authority (OGA). The performance of some of the most commonly used artificial intelligence methods (support vector machine, random forest, multi-layer perceptron) was tested varying their hyperparameters until reaching an adequate result.

Using Process Mining to Reduce Fraud in Digital Onboarding

In the context of online banking, new users have to register their information to become clients through mobile applications; this process is called digital onboarding. Fraudsters often commit identity fraud by impersonating other people to obtain access to banking services by using personal data obtained illegally and causing damage to the organisation’s reputation and resources. Detecting fraudulent users by their onboarding process is not a trivial task, as it is difficult to identify possible vulnerabilities in the process to be exploited. Furthermore, the modus operandi for differentiating the behaviour of fraudulent actors and legitimate users is unclear. In this work, we propose the usage of a process mining (PM) approach to detect identity fraud in digital onboarding using a real fintech event log. The proposed PM approach is capable of modelling the behaviour of users as they go through a digital onboarding process, while also providing insight into the process itself. The results of PM techniques and the machine learning classifiers showed a promising 80% accuracy rate in classifying users as fraudulent or legitimate. Furthermore, the application of process discovery in the event log dataset produced an insightful visual model of the onboarding process.

Experimental Analyses of Temporal Activity-Sequencing Anomalies in Process Mining

In the research field of the automated process discovery and analysis, the purity of event log datasets ought to be a matter of vital importance to the success of discovering sound and exact process models. Moreover, there exist various types of anomalies that result in the discovery of inaccurate process models from the process enactment event log datasets. A peculiar one out of these anomalies, which is the core challenging issue of this paper, is the temporal activity-sequencing anomaly that critically affects the overall quality of the automated process discovery. This paper explores such event-log anomalies and noises produced by the special type of anomalies inevitably formed in the event-log preprocessing phase of the automated process discovery. More precisely, it implements an algorithmic approach that is able to detect and filter out those anomalies and noises in performing the automated process discovery. The author also carries out a series of experimental analyses by applying the implemented approach to the five datasets of process event logs available in the 4TU Center for Research Data.

CE-SGAN: Classification enhancement semi-supervised generative adversarial network for lithology identification

Lithology identification, the process of recognizing and distinguishing lithology using specific methods, is a fundamental task in the fields of formation evaluation and reservoir description. However, the lithology identification accuracy of traditional models is insufficient because the distribution of small-sample logging data is usually extremely unbalanced. In this paper, a classification enhancement semi-supervised generative adversarial network (CE-SGAN) model is proposed to mitigate the influence of data imbalance and improve the lithology identification accuracy. Considering the powerful nonlinearity of logging curves, a classification separation architecture is applied to attain an optimal equilibrium between the classifier and the generator. Furthermore, a pseudo-label processing mechanism is designed to enhance classification, which combined with semi-supervised learning. Experiment results on two small sample logging datasets demonstrate that the model provides a considerable improvement in lithology identification. Moreover, it is competitive in data enhancement.

Oil Logging Reservoir Recognition Based on TCN and SA-BiLSTM Deep Learning Method

The use of Deep Learning methods to mine useful and critical information from massive and complex logging datasets is of great importance for oil logging reservoir recognition. TCN-SA-BiLSTM was proposed due to the lack of previous studies to mine the internal correlation of the features of the logging dataset. TCN-SA-BiLSTM is a deep learning model that hybridizes Temporal Convolutional Network (TCN), Self-Attention mechanism (SA), and Bidirectional Long Short Term Memory network (BiLSTM). First, for the pre-processed feature data, TCN is used for feature extraction with parallel convolution operation. Then, by exploiting the ability of SA to extract the internal autocorrelation of time series features, this can better capture the dependence of feature data over long distances. Finally, the contextual linkage of the features is further obtained using BiLSTM. The experimental results show that TCN-SA-BiLSTM exhibits excellent performance in comparison with seven competing models on all performance evaluation metrics. It overcomes the deficiencies in capability exhibited by traditional logging interpretation techniques to improve the efficiency and success rate of oil and gas exploration.

Channel attention-based static-dynamic graph convolutional network for lithology identification with scarce labels

Accurately identifying the complicated lithology distribution of unconventional reservoirs has become an urgent need for oil and gas exploitation. However, the expensive and rare lithology labels limit the precision of current machine learning-based lithology identification methods. Meanwhile, those methods neglect the associated information implicit in adjacent logging samples and underutilize the abundant unlabeled data. In this paper, a channel attention-based static-dynamic graph convolutional network for lithology identification is proposed to solve the problem of scarce labels. To utilize the unlabeled logging data adequately, logging graphs are modeled based on geographic distance and feature similarity to characterize associated information as well as logging features. On the basis, the dynamic similarity graph is introduced to magnify distinctions and reduce connections between samples with different lithologies, improving the quality of logging graphs. Aiming at enhancing the model performance of eliminating redundant and noisy information in logging graphs, the channel attention mechanism is taken into account to realize the fusion and complementation between different associated information. Verification experiments are conducted by utilizing the logging datasets from a shale formation. Results show that the proposed approach has general applicability and achieves high identification accuracy for various strata. Moreover, the proposed approach reduces the cost of drilling coring for lithology identification immensely and can serve as a reference for applying graph theory to solve future reservoir-related issues in petroleum exploitation.

Lithology identification using graph neural network in continental shale oil reservoirs: A case study in Mahu Sag, Junggar Basin, Western China

The continental shale oil reservoir of Fengcheng Formation in the northern slope area of Mahu Sag, Junggar Basin, Western China is very heterogeneous in lithology. Thus, the complex response characteristics of conventional logging and limited core availability in the study area has led to major challenges in lithology identification. Therefore, to resolve lithology identification by well logs in continental shale oil reservoirs, a graph neural network (GNN) method named GraphSAGE is used to train the lithology identification model based on a constructed graph, which connects the samples with adjacent depth and similar log response features on operator intention. The identification process is divided into two parts: first, based on the formation depth sequence and affinity propagation clustering method, the vertical distribution of the stratum and nodes logging curve similarity information are integrated into the graph structure, which structurally represents the conventional logging curves as graph instead of well logs as input data; Second, the nodes of the constructed graph are classified by GraphSAGE, which naturally supports combination generalization and improves sample complexity accompanied by strong relational inductive bias. To examine the effectiveness of GraphSAGE for lithology identification, a conventional log dataset labelled by direct core observations from two separate wells in Muhu Sag are used. The identification results showed that the accuracy of GraphSAGE for the lithologies exceeds 90% of the testing data, especially for transitional lithology such as dolomitic mudstone, silty mudstone and tuffaceous fine sandstone. Compared with the commonly used machine learning methods such as SVM, RF and XGBoost, GraphSAGE was more accurate in lithology identification, matching core observations. Collectively, this reflects the superiority of graph neural network in conventional logging lithology identification and effective means provided for lithology identification of continental shale oil reservoir in the Mahu Sag.

Characterization and Prediction of Mobile Tasks

Mobile devices have become an increasingly ubiquitous part of our everyday life. We use mobile services to perform a broad range of tasks (e.g., booking travel or conducting remote office work), leading to often lengthy interactions with several distinct apps and services. Existing mobile systems handle mostly simple user needs, where a single app is taken as the unit of interaction. To understand users’ expectations and to provide context-aware services, it is important to model users’ interactions with their performed task in mind. To provide a comprehensive picture of common mobile tasks, we first conduct a small-scale user study to understand annotated mobile tasks in-depth, while we demonstrate that by using a set of features (temporal, similarity, and log sequence), we can identify if a pair of app usage belong to the same task effectively. Secondly, the proposed best task detection model is applied to a large-scale data set of commercial mobile app usage logs to infer characteristics of complex (multi-app) mobile tasks in the wild. By applying an unsupervised learning framework, we discover common mobile task types that span multiple apps based on various extracted characteristics. We observe that users generally perform 17 common tasks with 47 sub-tasks, ranging from “social media browsing” to “dining out” and “family entertainments”. Finally, we demonstrate that we can predict the next complex mobile task that users are likely to perform by leveraging features from the historically inferred mobile tasks and user contexts. Our work facilitates an in-depth understanding of mobile tasks at scale, enabling applications for promoting task-aware services.

Delineating the main structural outlines and the petrophysical properties of the Albian-upper Cretaceous Reservoirs using seismic and well log data, Shushan Basin, north Western Desert, Egypt

This study concerns with delineating and analyzing the subsurface structural setting of Khalda Oil Field at Shushan Basin as a key factor for evaluating the petrophysical properties and the deliverability of the Bahariya Formation. Shushan Basin is considered one of the most prospective basins in the Western Desert. The oil is trapped within a gentle seismically defined anticline that contains a series of coastal sandstone reservoirs interbedded with a neritic and tidal flat mudstone and a few carbonate interbeds of the Albian-Cenomanian Bahariya Formation forming multiple oil-bearing sandstone reservoirs. A set of borehole logging and seismic data were available, tied, processed, and mapped to delineate the predominant structures in the study area. Results show the implication of the NW–SE, NE–SW, and E–W gravity faulting on the Bahariya and the overlying Abu Roash formations. This set of gravity faults is attributed to the Mesozoic tensional stresses. The depth structure contour maps on the upper surfaces of the Lower and Upper Bahariya members reveal the presence of some NE–SW gravity faults that dissected the Khalda Field into some slightly tilted blocks, shallowing in depth to the south as deepening to the north. Besides, the study area has been affected later in the Jurassic by E–W gravity faults during the Atlantic Ocean opening. This faulting trend was changed later into the NE–SW direction as a result of the movement of North Africa against Laurasia. The petrophysical analysis indicated that the Lower Bahariya sandstone reservoir has a good reservoir quality (16 ≤ ∅ ≤ 25%, 53 ≤ Sw ≤ 59%, 6.0 ≤ Vsh ≤ 32.2%, and good net-pay thickness 18.0–38.0 ft). It is revealed that the drilled wells penetrating the crest of the anticline are prospective while that located in the anticline flanks are water-bearing recommending exploration away from the anticline flanks. By applying this workflow, it is possible to explore for the similar subsurface hydrocarbons-bearing sequences in the Western Desert and North Africa in future exploitation plans.

Research on Volume Measurement of Logs Based on Embedded Application

Spurred by the worldwide concern for forest protection and the increased log sales, most countries have standardized log volume calculations to avoid excessive timber and protect buyers. However, log volume is currently manually measured, suffering from high labor costs, low measurement progress, and imposing significant measurement errors. Thus, automatically obtaining the volumetric data of logs is a convenient and quick solution. Therefore, this work proposes a Mask Region Convolutional Neural Network-based (R-CNN) algorithm for Logs volume measurement, named Wood Mask (WM) R-CNN. Specifically, we employ the Res2Net structure as the backbone to obtain receptive fields that exceed the input feature size, thus improving our model’s multi-scale information fusion ability. Additionally, WM R-CNN relies on the Path Aggregation Feature Pyramid Network’s (PAFPN’s) path enhancement structure, shortening the low-level feature map’s propagation path and improving the wood contour segmentation accuracy. Extensive experiments on the Vehicle-mounted Dense Logs (VMDL) dataset demonstrate that WM R-CNN affords a highly appealing segmentation accuracy for small, medium, and large wood, improving the corresponding mAP indicators against current methods by 2.0%, 1.2%, and 4.4%, respectively. Furthermore, a quantitative method based on TensorRT compresses the proposed model to deploy the WM R-CNN to mobile embedded devices. However, to compensate for the quantization loss, we introduce the expansion convolution operation method to manipulate the mask map and control the volume calculation error of all logs on a vehicle within 1%. The experiments reveal that the proposed method offers an appealing performance, verifying the algorithm’s effectiveness and implementation ability on mobile terminals.

Log Anomaly Detection Based on Hierarchical Graph Neural Network and Label Contrastive Coding

System logs are essential for detecting anomalies, querying faults, and tracing attacks. Because of the time-consuming and labor-intensive nature of manual system troubleshooting and anomaly detection, it cannot meet the actual needs. The implementation of automated log anomaly detection is a topic that demands urgent research. However, the prior work on processing log data is mainly one-dimensional and cannot profoundly learn the complex associations in log data. Meanwhile, there is a lack of attention to the utilization of log labels and usually relies on a large number of labels for detection. This paper proposes a novel and practical detection model named LCC-HGLog, the core of which is the conversion of log anomaly detection into a graph classification problem. Semantic temporal graphs (STG) are constructed by extracting the raw logs’ execution sequences and template semantics. Then a unique graph classifier is used to better comprehend each STG’s semantic, sequential, and structural features. The classification model is trained jointly by graph classification loss and label contrastive loss. While achieving discriminability at the class-level, it increases the fine-grained identification at the instance-level, thus achieving detection performance even with a small amount of labeled data. We have conducted numerous experiments on real log datasets, showing that the proposed model outperforms the baseline methods and obtains the best all-around performance. Moreover, the detection performance degrades to less than 1% when only 10% of the labeled data is used. With 200 labeled samples, we can achieve the same or better detection results than the baseline methods.

Petrophysical Evaluation of H-field, Niger Delta Basin for Petroleum Plays and Prospects

Abstract in English As a result of a combination of several methodologies, the H-Field, Niger-Delta, has been delineated. To identify probable reservoirs, seals, and source rocks in the study area, this study integrates sequence stratigraphy with petrophysical concepts through a comprehensive analysis of 3D seismic and well logging datasets. According to the 3D facies model, reservoirs are dominated by sand and are laterally extensive. They were then upscaled and stochastically distributed to create a 3D reservoir property model. On average, the porosity of these clastic reservoirs ranges from 22 to 28%. Reservoir net-to-gross (NTG) ratios range from 0.67 to 0.96. Water saturation ranges from 8% to 30%, while hydrocarbon saturation ranges from 70% to 92%. Four gas-bearing units and six oil-bearing units are present in reservoirs P 0.5 and P 1 compartments. All reservoir intervals’ oil and gas volumes are evaluated based on the hydrocarbon distribution. Geological models of the subsurface, such as the one in this paper, are crucial for future reservoir development programs.

Reservoir Characteristics of the Lower Miocene Carbonate Formations in Kor Mor Gasfield, Kirkuk Area, NE Iraq

This paper presents the evaluations of the reservoir characterization of the Lower Miocene Euphrates, and Jeribe carbonate formations that have been selected for this study in Kor Mor Gasfield in Northeastern Iraq. The Euphrates ranged in thickness between 30 and 65, and Jeribe 30 and 50m in this field, and they are overlain by a thick evaporate cap rock of the Fatha Formation. The two formations are separated by the thin evaporate rock unit of the Dhiban Formation
 The full set of log data for the well Km-9 was used in addition to some general data derived from other eight wells (Km-1 to Km-8). The log data have been used in evaluating the reservoir properties of the two rock units. The detected lithologies from the available mud log and well reports showed that the two formations mostly dolostone and dolomitic limestone were dominated. From the log data, different types of reservoir properties (clay content, clay types, porosity, and fracture index,) have been estimated. The average porosity of the Euphrates reservoir is around of 0.15, and the Jeribe 0.10, the secondary porosity makes a significant fraction of porosity in both units.

Interwell Stratigraphic Correlation Detection based on knowledge-enhanced few-shot learning

Interwell Stratigraphic Correlations Detection (ISCD) guides reservoir modeling and oil development. Many existing AI (artificial intelligence) methods have been proposed for ISCD. However, it is difficult to generate labels for large-scale geological data, which leads to the problem of small samples. In this paper, we propose a few-shot learning-based approach to detect stratigraphic correlations for overcoming this challenge. Specifically, we design a Knowledge Enhanced Few-shot Transformer ISCD model (KEFT-ISCD) to enhance reservoir sample features. We design a dynamically balanced marginal softmax (dbm-softmax) to further optimize the model loss for identifying edge features, which improves the stratigraphic matching effects. In addition, we design a bi-window co-sliding approach to address the cross-matching problem in practical stratigraphic matching. To the best of our knowledge, this is the first work to use few-shot learning for the ISCD. We evaluate the proposed method with different well sections in a pair of adjacent wells from a real-world well logging dataset. Experimental results indicate that the proposed KEFT-ISCD performs well and achieves a detection accuracy of 91.12%. We also conduct experiments on different wells and blocks. The results further demonstrate the generalizability of the proposed approach.

SiaLog: detecting anomalies in software execution logs using the siamese network

Detecting anomalies in software logs has become a notable concern for software engineers and maintainers as they represent anomalies in software execution paths and states. This paper propose a novel anomaly detection approach based on the Siamese network on top of Recurrent Neural Networks(RNN). Accordingly, we introduce a novel training pair generation algorithm to train the Siamese network which reduces generated training significantly while maintaining the F_1 score. Additionally, we propose a hybrid model by combining the Siamese network with a traditional feedforward neural network to make end-to-end training possible, reducing engineering effort in setting up a deep-learning-based log anomaly detector. Furthermore, we provides validations of the approach on the Hadoop Distributed File System (HDFS), Blue Gene/L (BGL), and Hadoop map-reduce task log datasets. To the best of our knowledge, the proposed approach outperforms other methods on the same dataset at the F_1 scores of respectively 0.99, 0.99, and 0.94 on HDFS, BGL, and Hadoop datasets, resulting in a new state-of-the-art performance.To further evaluate the proposed method, we examine our method’s robustness to log evolutions by evaluating the model on synthetically evolved log sequences; we got the F_1 score of 0.95 on the HDFS dataset at the noise ratio of 20%. Finally, we dive deep into some of the side benefits of the Siamese network. Accordingly, we introduce an unsupervised log evolution monitoring method alongside a visualization technique that facilitates model interpretability.

Machine invasion: Automation in information acquisition and the cross-section of stock returns

We estimate the number of machines “covering” a firm by separating machine Internet protocols (IPs) from human IPs based on the intensity of information retrieval using the EDGAR web log dataset. We investigate the relationship of machine coverage and the cross-section of stock returns and find that stocks in the lowest quintile of machine coverage outperform those in the highest quintile by 6% annually after adjusting for risk. Our results indicate that automation in information processing has a significant impact on the cross-section of stock returns.

Network anomaly detection based on keyword embedding log

Log anomaly detection is an important and challenging task in the field of Artificial Intelligence for IT Operations (AIOps). Logs that record important runtime information are widely used for troubleshooting purposes. There have been many studies that use log data to construct deep learning methods for detecting system anomalies, which are usually based on log parsing. However, they ignore the effect of keywords that are promising for system status analysis. Here, we propose KELog (Keyword Embedding Log), a novel log anomaly detection approach that utilizes keyword information. We build a keyword library by keyword information extraction and fuse them into log representations. In this way, KELog can raise the reliability of anomaly detection. The experimental results on a real-world log dataset of a communications operator show that the F1 score of our proposed KELog method achieves a maximum increase of 0.341 compared with the commonly used machine learning algorithms (PCA, SVM, Invaiant Mining) and a maximum increase of 0.039 compared with deep learning algorithms (DeepLog, LogBERT) respectively. In 2021, ITU launched the second ITU AI/ML in 5G Challenge. We used KELog to participate in the thematic track of the Artificial Intelligence Innovation and Application Competition in the China Division, and won first place with a full F1 score.

LRU-GENACO: A Hybrid Cached Data Optimization Based on the Least Used Method Improved Using Ant Colony and Genetic Algorithms

An optimization strategy for cached data offloading plays a crucial role in the edge network environment. This strategy can improve the performance of edge nodes with limited cache memory to serve data service requests from user terminals. The main challenge that must be solved in optimizing cached data offloading is assessing and selecting the cached data with the highest profit to be stored in the cache memory. Selecting the appropriate cached data can improve the utility of memory space to increase HR and reduce LSR. In this paper, we model the cached data offloading optimization strategy as the classic optimization KP01. The cached data offloading optimization strategy is then improved using a hybrid approach of three algorithms: LRU, ACO, and GA, called LRU-GENACO. The proposed LRU-GENACO was tested using four real proxy log datasets from IRCache. The simulation results show that the proposed LRU-GENACO hit ratio is superior to the LRU GDS SIZE algorithms by 13.1%, 26.96%, 53.78%, and 81.69%, respectively. The proposed LRU-GENACO method also reduces the average latency by 25.27%.