Embodiments of the invention are directed to active detection and mitigation of anomalies in program code construction interfaces. The system provides a proactive plug-in with a dynamic machine learning (ML) anomaly detection model cloud component structured to dynamically detect architectural flaws in program code in real-time in a user coding interface. In particular, the system activates a machine learning (ML) anomaly detection plug-in for dynamically analyzing the first technology program code being constructed in the user coding interface. Moreover, the system modifies, via the ML anomaly detection plug-in, the user coding interface to embed interface elements associated with the one or more flaws in the first technology program code detected by the ML anomaly detection model cloud component.

A conversion apparatus improves utilization of definition information for system migration by including: a first conversion unit converting definition information about first rules for generating first data showing information about an operation of a migration source first system in a form depending on an implementation technique of the first system based on materials of the first system, according to a combination of first environment information showing an environment that the first system depends on and second environment information showing an environment that a migration destination second system depends on; a second conversion unit converting definition information about second rules for generating second data showing the information about the operation in a form not depending on a particular implementation technique based on the first data, according to the combination of the first environment information and the second environment information; and a third conversion unit converting definition information about third rules for generating third data showing the information about the operation in a form depending on an implementation technique of the second system based on the second data and the first data, according to the combination of the first environment information and the second environment information.

To understand/reverse engineer the code, knowledge of cryptic terms (variable names) present in the code is mandatory. The reverse engineering to understand the code is a very complex task which has infinite variations. The present disclosure provides a method and system for identifying meaningful terms in a domain context from a plurality of cryptic forms of a variable name in a program code. The present disclosure provides a machine learning model that understands the cryptic form of a variable name and relates the co-occurring cryptic terms and expands them. These expanded forms of cryptic terms directly aid in understanding of each term and its usage in a more accurate way. This knowledge is used in many downstream task of reverse engineering the program code. This disclosure links the multiple usages of the same variable and aims to reduce the gap of naming convention mismatches introduced by developers.

To understand/reverse engineer the code, knowledge of cryptic terms (variable names) present in the code is mandatory. The reverse engineering to understand the code is a very complex task which has infinite variations. The present disclosure provides a method and system for identifying meaningful terms in a domain context from a plurality of cryptic forms of a variable name in a program code. The present disclosure provides a machine learning model that understands the cryptic form of a variable name and relates the co-occurring cryptic terms and expands them. These expanded forms of cryptic terms directly aid in understanding of each term and its usage in a more accurate way. This knowledge is used in many downstream task of reverse engineering the program code. This disclosure links the multiple usages of the same variable and aims to reduce the gap of naming convention mismatches introduced by developers.

A neural network model conversion method, a server, and a storage medium are provided according to embodiments of the present disclosure. The neural network model conversion method includes: parsing a neural network model to obtain initial model information; reconstructing the initial model information to obtain streaming model information; generating a target model information file according to the streaming model information; and running, under a streaming architecture, the neural network model according to the target model information file.

A neural network model conversion method, a server, and a storage medium are provided according to embodiments of the present disclosure. The neural network model conversion method includes: parsing a neural network model to obtain initial model information; reconstructing the initial model information to obtain streaming model information; generating a target model information file according to the streaming model information; and running, under a streaming architecture, the neural network model according to the target model information file.

Source code is analyzed to identify components. The components are each assigned a complexity score. Documentation for the source code is identified, related to the components, and given a score based on the quantity of the documentation for the component and the complexity score for the component. To determine semantic meaning of the documentation, vector embeddings for the documentation languages may be generated and aligned. Alignment causes the different machine learning models to generate similar vectors for semantically similar words in the different languages. Since the vectors of the words of the other languages are similar to the vectors of the words in a primary language with similar meanings, the vector representation of the documentation in the other languages will match the vector representation of the source code when the documentation is substantially on the same topic.

Source code is analyzed to identify components. The components are each assigned a complexity score. Documentation for the source code is identified, related to the components, and given a score based on the quantity of the documentation for the component and the complexity score for the component. To determine semantic meaning of the documentation, vector embeddings for the documentation languages may be generated and aligned. Alignment causes the different machine learning models to generate similar vectors for semantically similar words in the different languages. Since the vectors of the words of the other languages are similar to the vectors of the words in a primary language with similar meanings, the vector representation of the documentation in the other languages will match the vector representation of the source code when the documentation is substantially on the same topic.

Some embodiments of the teachings herein include a computer-implemented method for automatic derivation of attributes of software engineering artifacts arising from technical boundary condition of product or service development segments comprise: deducing technical requirements using an automated software-based process based on classifications of the technical boundary conditions; mapping the deduced technical requirements of the artifacts to engineering disciplines and concerns using an automated software-based process; mapping the calculated engineering artifacts to responsibilities; adapting the classification of the technical boundary conditions based on the evaluation results in iterations; and processing an executable performing a distribution calculation of the classification space, wherein the distribution calculation of the classification space is at least based on a distribution and quartiles. The process of the executable comprises: calculating combination vectors at system start; reading mapping data and calculation probabilistic distribution and quartiles; and publishing new distribution to engineering goal calculation.

A difference extracting unit extracts, from an executable file converted from a source code and an executable file converted from a source code after vulnerability correction is made to the source code, a difference of a part where the vulnerability correction is made. A feature calculating unit calculates features of the difference extracted by the difference extracting unit. A difference extracting unit extracts, from an executable file converted from a source code and an executable file converted from a source code after correction is made to the source code, a difference of a predetermined part. A similarity calculating unit calculates similarity between the difference of the predetermined part calculated by the difference extracting unit and the features of the difference of the part where the vulnerability correction is made calculated by the feature calculating unit.