Abstract: The disclosure relates to a method and a system for collecting and processing offensive and defensive elements based on multi-source data. The method includes: constructing multiple data monitoring points based on network system parameters; generating a collection evaluation value of each of the data monitoring points, and setting an initial collection frequency of each of the data monitoring points according to all collection evaluation values; obtaining a monitoring data packet of each of the data monitoring points, and determining whether to generate a first-level collection instruction of each of the monitoring points according to all monitoring data packets.

Abstract: Provided is a method for constructing a backtracking analysis model based on an attack chain, including: a kernel-layer security agent collects process, file, and network behavior features in a hardware-isolated environment to generate event tuples; a tensor network pipeline performs three-dimensional decoupling mapping of the tuples into behavior fingerprint vectors, orthogonalized noise features, and asymmetric adjacency tensors, and compresses them into spatio-temporal topological tensor block; a reinforcement learning controller constructs a directed acyclic graph based on the tensor block, calculates connectivity loss, and outputs event risk scores; a dynamic routing engine constructs a decision tree model based on risk markers, burst frequency, and correlation entropy to implement three-level shunting, and a multi-simulation system generates anti-interference index; when the deviation between a physical trajectory and a digital model exceeds a tolerance, a closed-loop feedback weight coefficient is used

Abstract: Provided are a distributed task load balancing scheduling method and system. The method includes: acquiring performance requirement indicators of tasks to be executed, and allocating the tasks to be executed to server nodes in a server cluster network topology according to the performance requirement indicators; acquiring historical load change rates of the server nodes, and determining a load threshold corresponding to each server node according to the historical load change rates; and acquiring real-time performance indicators of the server nodes, determining load score values according to the real-time performance indicators of the server nodes, screening out overloaded server nodes whose load score values exceed the load thresholds, and performing load balancing scheduling on tasks in the overloaded server nodes.

Abstract: Techniques for creating an interest graph include obtaining content items from multiple content sources and applying tailored (e.g., source-specific) preprocessing to the content items based on their respective content source. Text is extracted and salient keywords and key phrases are identified using unsupervised machine learning models. The keywords and key phrases become nodes in an interest graph, each node comprising an embedding of a keyword or key phrase in a common embedding space, with edges representing semantic similarity based on embeddings or co-engagement patterns. The graph provides an expansive, granular, and dynamic taxonomy easily adaptable to emerging interests. The interest graph overcomes limitations of conventional taxonomies that lack depth, fail to capture niche interests, and cannot adapt to reflect evolving user preferences. The described techniques construct a rich interest graph from diverse content for improved content understanding.

Abstract: The disclosure belongs to the technical field of network security, and provides a method for constructing a feature knowledge base of mapping behavior based on deep learning, which includes: data acquisition and preprocessing: extracting five-tuple information and behavior features from network traffic. The disclosure automatically extracts the spatio-temporal features through the deep learning model, and enhances the sensitivity to abnormal behaviors by combining the attention mechanism, thus significantly improving the detection accuracy. The explanatory AI technology is used to automatically generate detection rules, the maintenance cost of manual rules is greatly reduced and the efficiency of rule generation is significantly improved. The feature knowledge base supports dynamic updating, may integrate third-party threat information in real time, and ensures the continuous defense ability against new attacks and variant detection means.

Abstract: The disclosure relates to the technical field of cloud resource pools, and in particular to a cloud security source pool system based on distributed architecture. The system includes: a central control unit, configured for building a cloud resource pool according to security device parameters, where the cloud resource pool includes multiple virtual security machines; and a security monitoring unit, configured for setting multiple request categories according to network structure parameters. The security monitoring unit is further configured for constructing a security sub-model of each of the request categories. Based on the distributed architecture and virtualization technology, each security device is built as a virtual security machine, and a cloud resource pool is built according to all virtual security machines, so as to realize the dynamic call of all security resources, and at the same time, multiple request categories are built based on historical parameters.

Abstract: A network mapping behavior anomaly detection method and system based on machine learning is provided. The method includes: collecting dual-source traffic data, generating a structured log data set through dual-source log fusion engine; performing subgraph matching calculation to obtain a mapping behavior deviation degree; generating communication data containing a watermark identifier in a session corresponding communication path; verifying whether attack events carry the watermark identifier; generating a network mapping behavior anomaly detection report.

Abstract: Provided are harmless attack verification and analysis method and system. The method includes: acquiring multi-source security data, determining attack cases in the data to construct an attack knowledge graph; constructing a scenario-based attack flow chart based on the graph, and formulating a harmless attack task based on the chart; determining a number of paths and a number of nodes of attack paths in the harmless attack task, and decomposing the harmless attack task into sub-tasks based on the number of paths and the number of nodes; acquiring path information of each attack path, determining a complexity of the attack path based on the path information, and determining attack priorities of the sub-tasks based on the complexities of the attack paths; and determining an attack sequence of the sub-tasks based on the attack priorities, and performing simulated attack verification on a protection boundary by the sub-tasks in the attack sequence.

Abstract: A network security operation workbench is provided, and relates to the technical field of network security, and includes: a data monitoring module, configured for performing security monitoring on network events in systems, identifying abnormal data and outputting to obtain abnormal information; a threat analysis module, configured for performing secondary identification on the abnormal information by using preset identification algorithms, obtaining threat types and levels corresponding to abnormal events in the abnormal information, and generating threat identification information; a risk processing module, configured for analyzing the threat identification information, matching to obtain corresponding coping strategies and methods, and performing security management operations on the abnormal events based on the coping strategies and methods; a log generation module, configured for recording an identification and analysis process and a security management operation process of each of abnormal events in the

Abstract: Techniques for creating an interest graph include obtaining content items from multiple content sources and applying tailored (e.g., source-specific) preprocessing to the content items based on their respective content source. Text is extracted and salient keywords and key phrases are identified using unsupervised machine learning models. The keywords and key phrases become nodes in an interest graph, each node comprising an embedding of a keyword or key phrase in a common embedding space, with edges representing semantic similarity based on embeddings or co-engagement patterns. The graph provides an expansive, granular, and dynamic taxonomy easily adaptable to emerging interests. The interest graph overcomes limitations of conventional taxonomies that lack depth, fail to capture niche interests, and cannot adapt to reflect evolving user preferences. The described techniques construct a rich interest graph from diverse content for improved content understanding.

Abstract: A method for generating an operation scheduling scheme of a hydrogen-photovoltaic-storage-charging integrated energy station includes the following steps: setting operating parameters of a hydrogen-photovoltaic-storage-charging integrated energy station; establishing an operation scheduling model of the hydrogen-photovoltaic-storage-charging integrated energy station according to the operating parameters; and finally, iteratively solving the operation scheduling model of the hydrogen-photovoltaic-storage-charging integrated energy station by means of an improved grey wolf optimization algorithm to obtain individual position with a maximum predation benefit in a current wolf population, and outputting the individual position as an optimal scheduling scheme of the hydrogen-photovoltaic-storage-charging integrated energy station.

Abstract: Techniques for creating an interest graph include obtaining content items from multiple content sources and applying tailored (e.g., source-specific) preprocessing to the content items based on their respective content source. Text is extracted and salient keywords and key phrases are identified using unsupervised machine learning models. The keywords and key phrases become nodes in an interest graph, each node comprising an embedding of a keyword or key phrase in a common embedding space, with edges representing semantic similarity based on embeddings or co-engagement patterns. The graph provides an expansive, granular, and dynamic taxonomy easily adaptable to emerging interests. The interest graph overcomes limitations of conventional taxonomies that lack depth, fail to capture niche interests, and cannot adapt to reflect evolving user preferences. The described techniques construct a rich interest graph from diverse content for improved content understanding.

Abstract: Techniques for automated tagging of visual content are described. A pairwise model is used to encode images and videos with a vision encoder, and encode keywords and phrases from an interest graph with a text encoder. Similarity layers compare these cross-modality embeddings by calculating distance in a shared embedding space. Scores indicate the association between visual features and text. A pairwise loss function brings together matched pairs while separating non-matches during training. Scores exceeding a threshold tag content with relevant keywords and phrases. The pairwise architecture relates images and text despite limited associated text. It leverages vision-to-text understanding for accurate tagging without per-class labels. Contrastive similarity techniques associate visual patterns with textual concepts. Automated tagging organizes user-generated content by topics using this scalable cross-modality approach.

Abstract: An embodiment of the present disclosure provides a leveling agent, which is specifically a polypyridine compound, where the polypyridine compound includes a structural unit shown in Formula (I) or a protonated product of the structural unit shown in Formula (I): In Formula (I), R1 and R2 are independently any one of the following: substituted or unsubstituted alkylene, substituted or unsubstituted arylene, substituted or unsubstituted arylene alkyl, substituted or unsubstituted alkylene aryl, alkylene containing an ether oxygen atom, ester, and/or imide, arylene containing an ether oxygen atom, ester, and/or imide, arylene alkyl containing an ether oxygen atom, ester, and/or imide, and alkylene aryl containing an ether oxygen atom, ester, and/or imide; and R3 is any one of groups such as a single bond, substituted or unsubstituted alkylene, substituted or unsubstituted arylene, substituted or unsubstituted arylene alkyl, and substituted or unsubstituted alkylene aryl.

Abstract: The invention discloses a multi-stage shock absorption and seismic resistance system for urban underground comprehensive pipe tunnels. It includes a pipe gallery tunnel and pipeline grooves. Inside the tunnel, a multi-segment active cavity with multiple arc-shaped inner cavity walls forms a pipe shape with the same axis line as the tunnel. Each inner cavity wall has a sealing elastic contact pad, ensuring a tight seal. Adjacent active cavities connect end to end, forming an inner tunnel body that extends with the pipe tunnel. The outer wall of the inner cavity is fixed to the tunnel's inner wall by a fixed frame. The pipeline groove is installed on the inner wall of the inner cavity through a telescopic frame. The inner tunnel body can swing up and down or left and right, absorbing transverse and longitudinal earthquake waves and resisting impact and collapse.

Abstract: A chatbot system detects commercial intent during conversations with users and provides targeted advertisements. The chatbot system extracts keyword candidates from a conversation and assigns relevance scores based on the meaning of the conversation and commercial scores based on a machine learning model trained to detect commercially-related keywords. The chatbot system selects keywords based on a combination of the scores and transmits the selected keywords to advertising content servers that provide advertising content including advertisements selected using the keywords. The chatbot system displays the advertisements to the user during the conversation.

Abstract: The invention discloses a system for enhancing the damping and aseismic capacities of an underground integrated pipe rack. It includes a pipe fixing assembly installed on the inner walls of the underground space on both sides. The assembly comprises multiple lifting frames arranged vertically and a lifting mechanism. The telescopic end of the lifting mechanism is fixedly connected to the lowermost lifting frame. One side of the lifting frame is slidingly connected to the inner wall of the underground space and features a recess for placing the pipe. A pipe holder inside the recess secures the pipe. During vibrations, the lifting frame slides on the inner wall, causing relative movement between adjacent pipe holders. The vibration is absorbed by the telescopic spring, preventing hard contact between the pipe and the vibration source. This reduces vibration and protects the pipe, thereby enhancing the aseismic capacity of the underground pipe rack.

Abstract: A transaction processing method includes: determining a first moment from a query time period of a target transaction, the first moment being a moment at which a most recent dumping of historical-state data is completed, and the historical-state data meeting a data query condition of the target transaction; obtaining a full analysis result of the historical-state data by processing the historical-state data based on analysis semantic of the target transaction; obtaining an incremental analysis result based on a specific change that occurs on current-state data corresponding to the historical-state data, the specific change occurring between the first moment and an end moment of the query time period, and the incremental analysis result being obtained by processing, based on the analysis semantic, data obtained by the specific change; and obtaining a final analysis result of the target transaction based on the full analysis result and the incremental analysis result.

Abstract: The disclosure provides a security analysis method and system based on protocol state, which relates to the technical field of protocol security protection. The method includes the following: a node traversal table is built, the node traversal table is scanned and analyzed according to the protocol trigger sequence rule, a first security evaluation factor of a protocol stack is determined, and a second security evaluation factor of each protocol is determined based on protocol normal application rule, and the trustworthiness degree of the second security factor is determined based on the first security factor, and the second security factor is revised based on the trustworthiness degree, and the security state of the protocol is determined according to the revised second security factor, thus the analysis of the protocol state is realized, and the security of the protocol can be accurately determined.

Abstract: The present invention provides a novel groove cutter with chip breaker, including: a cutter body; a cutter head arranged at the end part of the cutter body, a cutting edge being arranged at a bordering position of the cutter head; and a chip breaker arranged on the cutter head, wherein the chip breaker comprises a breaker bottom and a breaker wall, the breaker bottom being connected to the cutting edge, the breaker wall being arranged on one side of the breaker bottom away from the cutting edge, the breaker bottom being non-coplanar with the breaker wall, an intersecting line of the breaker bottom and the breaker wall including multiple connecting sections, and one curved transition section being arranged between two adjacent connecting sections. By means of the technical solutions provided in the present application, the problem that breaking chips in a groove machining process is difficult in the prior art can be solved.