Abstract:
PROBLEM TO BE SOLVED: To provide a parsing device and method, such as an XML parser, that quickly parse computerized structured documents. SOLUTION: State transitions of a plurality of states into which structured documents can be divided are integrated to generate optimum automatons according to statistical information and schema information about instance documents. In the state transition integration, continuously matching state transitions are held in ID lists, and the continuous state transitions are counted with the ID lists. For repetitive elements including nests, patterns of appearance frequency are statistically identified. XML blank variations are also statistically accommodated. The advance automaton construction further using the schema information reduces initialization overheads of the parsing device. COPYRIGHT: (C)2007,JPO&INPIT
Abstract:
PROBLEM TO BE SOLVED: To efficiently marge the change of a target model due to updating of a source model and the independent change of the target model. SOLUTION: In this model edit device 10, a model conversion function 11 converts an SM(source model)0 into a TM(target model)0, and generates a TDM(change dependence model)0. Also, when an SM editor generates an SM1 by updating the SM0, a model conversion function 11 generates a TDM1 by converting the SM1 into a TM1. When a TM editor generates TM0_n by editing the TM0, a Change element registration means 12 generates TDM0_n by registering a difference Δ2 between the TM0 and the TM0_n in the TDM0. Afterwards, the marge function 13 merges a difference Δ1 between the TDM0 and the TDM1 extracted by a Δ1 extraction function 14 and the difference Δ2 extracted by the TDM0_n based on a preliminarily registered processing pattern in response to calling. COPYRIGHT: (C)2009,JPO&INPIT
Abstract:
PROBLEM TO BE SOLVED: To improve performance of operation of a web application server. SOLUTION: A server 100 for receiving an HTTP request and generating an HTML file includes: an application execution part 120 for generating the HTML file; and an HTTP server 110 for transmitting an HTTP response including the HTML file in response to the HTTP request. The application execution part 120 includes a means for executing a script on generating the HTML file, and when an instruction code of the script is an instruction for processing a character string, calculating the number of appearances of the character string, and storing the character string whose number of appearances is a fixed value and more in a storage device 140 as a character string file. The application execution part 120 further includes a means which, when the character string coincides with the stored character string, transmits a character string file specified by reference information together with the HTML file. COPYRIGHT: (C)2011,JPO&INPIT
Abstract:
PROBLEM TO BE SOLVED: To efficiently retrieve a function having a desired function. SOLUTION: This system is a system for supporting retrieval of a function and includes a first storage device for storing history of at least one output parameter output from a plurality of functions while respectively associated with the functions, a parameter input part for receiving an input of an output parameter expected to be output from a function to be retrieved, a calculation part for calculating a first index-value showing the closeness to a value between the input output parameter and the respective output parameters stored in the first storage device about the respective output parameters stored in the first storage device and also totalizing the calculated first index-values for each function, and an output part for selecting and outputting functions, the totalized first index-values of which are larger than a predetermined reference value among the plurality of functions. COPYRIGHT: (C)2009,JPO&INPIT
Abstract:
PROBLEM TO BE SOLVED: To easily generate mapping between each property included in an object in a program and a structured document. SOLUTION: The device for generating mapping between each property included in an object in a program and each element of a structured document includes: an object tree generation part for assigning each property included in the object as a node of a tree structure, and for generating a tree structure showing the hierarchical structure of the object; and a selection part for selecting the mapping to minimize conversion costs for converting the tree structure of the object into a tree structure using each of the elements of the structured document as nodes from among mapping for associating each property included in the object with each of the elements of the structured document. COPYRIGHT: (C)2010,JPO&INPIT
Abstract:
PROBLEM TO BE SOLVED: To improve the performance of the operation of a Web application by reducing processing costs due to memory copy. SOLUTION: A server includes an application execution part 120 for dynamically generating an HTML file; and an HTTP server 110 for receiving an HTTP request, and for returning an HTTP response including an HTML file generated by the application execution part 120. The application execution part 120 executes the script of the HTML file, and when the instruction code of the script is to display the contents of a prescribed file stored in a storage device 140, the application execution part describes information to specify the file in an object without reading the file. The HTTP server 110 specifies the file stored in the storage device 140 on the basis of this information, and transmits the file with the HTML file. COPYRIGHT: (C)2010,JPO&INPIT
Abstract:
Bereitgestellt werden Techniken, die eine Erzeugung eines verschmolzenen Kernels ermöglichen, der einen vollen Kernel eines neuronalen Faltungsnetzes approximieren kann. In einem Beispiel umfasst ein durch einen Computer implementiertes Verfahren Bestimmen eines ersten Musters aus Abtastwerten einer ersten Abtastwertmatrix und eines zweiten Musters aus Abtastwerten einer zweiten Abtastwertmatrix. Die erste Abtastwertmatrix kann einen spärlich besetzten Kernel repräsentieren, und die zweite Abtastwertmatrix kann einen komplementären Kernel repräsentieren. Das erste Muster und das zweite Muster können zueinander komplementär sein. Das durch einen Computer implementierte Verfahren umfasst zudem Erzeugen eines verschmolzenen Kernels auf Grundlage einer Kombination aus Merkmalen des spärlich besetzten Kernels und Merkmalen des komplementären Kernels, die gemäß einem Verschmelzungsansatz kombiniert werden, und Trainieren des verschmolzenen Kernels.