Figure - uploaded by Antonio Mastropaolo
Content may be subject to copyright.
Source publication
Logging is a practice widely adopted in several phases of the software lifecycle. For example, during software development log statements allow engineers to verify and debug the system by exposing fine-grained information of the running software. While the benefits of logging are undisputed, taking proper decisions about where to inject log stateme...
Contexts in source publication
Context 1
... we remove duplicate methods from the dataset to avoid overlap between training and test sets we built from them. This left us with 6,909,280 methods, that we used to create the pre-training and the fine-tuning datasets summarized in Table 1. We used all the methods not having log statement (w/a log in Table 1) to build the dataset needed for the pre-training Task #P1 (i.e., the denoising task in which we randomly mask 15% of tokens). ...Context 2
... left us with 6,909,280 methods, that we used to create the pre-training and the fine-tuning datasets summarized in Table 1. We used all the methods not having log statement (w/a log in Table 1) to build the dataset needed for the pre-training Task #P1 (i.e., the denoising task in which we randomly mask 15% of tokens). For the pre-training Task #P2 (i.e., guessing the correct position of a log statement), we used 50% of methods with log statements: A method featuring í µí± log statements is present í µí± times in the dataset, each time with a different log statement removed. ...Context 3
... answer RQ 1 and RQ 2 we run against the test set ( Table 1) the best-performing configuration (Section 2.3) of the four models output of the different pre-trainings. Then, we assess the accuracy of the predictions generated by each model. ...Context 4
... third scenario (i.e., all three "components" are correct) is represented by the previously discussed correct predictions. The number of instances in the test set is 12,020, where each instance represents one of the 7,125 Java methods in Table 1 with a specific log statement removed (one method can have multiple log statements). ...Context 5
... we remove duplicate methods from the dataset to avoid overlap between training and test sets we built from them. This left us with 6,909,280 methods, that we used to create the pre-training and the fine-tuning datasets summarized in Table 1. We used all the methods not having log statement (w/a log in Table 1) to build the dataset needed for the pre-training Task #P1 (i.e., the denoising task in which we randomly mask 15% of tokens). ...Context 6
... left us with 6,909,280 methods, that we used to create the pre-training and the fine-tuning datasets summarized in Table 1. We used all the methods not having log statement (w/a log in Table 1) to build the dataset needed for the pre-training Task #P1 (i.e., the denoising task in which we randomly mask 15% of tokens). For the pre-training Task #P2 (i.e., guessing the correct position of a log statement), we used 50% of methods with log statements: A method featuring í µí± log statements is present í µí± times in the dataset, each time with a different log statement removed. ...Context 7
... answer RQ 1 and RQ 2 we run against the test set ( Table 1) the best-performing configuration (Section 2.3) of the four models output of the different pre-trainings. Then, we assess the accuracy of the predictions generated by each model. ...Context 8
... third scenario (i.e., all three "components" are correct) is represented by the previously discussed correct predictions. The number of instances in the test set is 12,020, where each instance represents one of the 7,125 Java methods in Table 1 with a specific log statement removed (one method can have multiple log statements). ...Similar publications
Developing a software service requires a strict software development life cycle and process. This process demands controlling all application code through source control management as well as a rigorous versioning and branching strategy. However, the platform and infrastructure also benefit from this rigor. Software services must be deployed to a t...