TWI427494B - A patent document search system, processing method, and search method with cloud structure - Google Patents

Description

Patent document retrieval platform, processing method and retrieval method thereof for cloud architecture

The invention provides a patent document retrieval platform, a processing method and a retrieval method thereof for a cloud architecture, in particular, a retrieval platform for applying a patent document, and uses a distributed storage location architecture in the retrieval platform to perform a fast retrieval operation.

With the increasing emphasis on intangible assets such as intellectual property rights, more and more enterprises or individuals apply for patents that they have developed to become patents, so as to obtain the protection of patent rights and exclude others from improper use of their research and development results. Under such a trend, the number of patent applications in various countries has gradually increased, and the capacity of the database for storing patent documents has become larger.

However, many companies need to research patent documents through some professional search platforms, perform analysis and statistics of patent documents to obtain future research and development directions, and make analysis and statistics difficult with the increasing number of patent documents.

Under the current analysis and statistical implementation of the patent document retrieval platform, it is usually necessary to regularly obtain patent documents from official patent file storage databases of various countries, and then compare them with all patent documents stored in the patent software according to the user's search conditions. Correct. In the case of such an increase in the number of patent documents, the search speed becomes slower.

Furthermore, the current patent document retrieval platform can only perform cross-statistical analysis on the number of patent documents, inventors, applicants, citation relationships, etc., and relatively, the content of the patent specification and the scope of application rights cannot be deeply Extracting more important and interrelated information from the middle, that is, according to the prior art, if the enterprise or the general user wants to know the technical similarity or relationship between the technical documents and the patent documents in the database, To become a clear study of the detailed research or research and development direction of litigation, you must read the multiple patent documents in the database one by one, and use the professional knowledge and experience to judge the relevance of the two through human judgment. However, with the development of science and technology, more and more patent materials are created every day, and the technical content involved is gradually complicated, making it more and more difficult to use patents to judge patent relevance, and it is becoming more and more economical. . If the research or R&D direction of these litigations is based entirely on human judgment, and the important decisions are made with the help of scientific methods, the current patent document retrieval platform will fall into the category. A lack of superficiality and excessive variability.

Therefore, the above-mentioned patent search platform still has many problems and deficiencies in the actual operation, which is the goal that the inventors and those engaged in the industry desire to improve.

According to the embodiment of the present invention, a patent document retrieval platform, a processing method, and a retrieval method of the cloud architecture can be provided. Through the different attributes of patent document content as the benchmark of the decentralized architecture, and using data analysis and other analytical methods, more and more patent documents can be quickly retrieved and analyzed.

In one embodiment, the disclosed method relates to a method of processing a patent document retrieval platform of a cloud architecture. This includes extracting a plurality of patent documents; parsing the plural attribute contents of the patent documents, respectively generating the same attribute content and different attribute contents; transmitting the same and different attribute contents to the plural storage locations; making the same attribute The content is stored in the same storage location respectively, and the different attribute contents are respectively stored in different storage locations, and then the retrieval request is received, and the comparison retrieval request is exactly the same as the attribute content in each storage location. Or partially identical, generate a search result, and respond to the search interface.

In another embodiment, the disclosed method is a retrieval method for a patent document retrieval platform of a cloud architecture. The method includes providing a search interface for receiving a search request; inputting a search request to the search interface; determining whether the search request includes a plurality of search conditions; and if there is a complex search condition, coupling the complex search condition, generating a search parameter, and then executing The comparison program; if there is a single search condition, the comparison program is executed; the matching is identical or partially identical to the attribute content; the search result is generated and the search result is returned to the search interface.

In yet another embodiment, the disclosed method is a retrieval method for a patent document retrieval platform of a cloud architecture. The method includes providing a search interface for receiving a search request; inputting a search request to the search interface; determining whether the search request includes a plurality of search conditions; and if there is a complex search condition, coupling the complex search condition, generating a search parameter, and then executing Comparing the program; if there is a single search condition, executing the comparison program; and obtaining the search result after the comparison meets the identical or partially identical attribute content; mining the same or partially identical to the at least one search condition The correlation between the content of the complex attribute and the content of the complex storage location; generating a data mining result and responding to the data mining result to the search interface.

In yet another embodiment, the disclosed method provides a search interface for receiving a search request for a patent document retrieval platform of a cloud architecture; inputting a search request to a search interface, wherein the search request is at least one technical document or patent document Parsing the at least one technical file or patent file, and generating a plurality of identical and different attribute contents; determining whether the search request includes a plural search condition; if there is a complex search condition, coupling the complex search condition and generating a search parameter, Then, the comparison program is executed; if there is a single search condition, the comparison program is executed; after the comparison, the attribute content that meets the exact same or partially the same is generated to generate a search result; and the search result is returned to the search interface.

In still another embodiment, the disclosed method provides a search interface for receiving a search request for a patent document retrieval platform of a cloud architecture; coupling and searching at least one combination of at least one search condition, and generating a word search parameter; Whether the word retrieval parameter and the attribute content of the plural storage location are identical or partially identical; the matching attribute content that meets the exact same or partially identical content is generated to generate a retrieval result; and the retrieval result is returned to the retrieval interface.

In order to achieve the above objects and effects, the technical means and the configuration of the present invention will be described in detail with reference to the preferred embodiments of the present invention.

Please refer to the first and second figures, which are flowcharts of processing and platform architecture of a preferred embodiment of the present invention. It can be clearly seen from the figure that the disclosed embodiment is related to a retrieval platform. 1. The search platform 1 is connected to the remote system 2 and the search interface 3 respectively, so that the search platform 1 can retrieve the plurality of patent documents stored by the remote system 2, and store the patent documents in the search platform 1 respectively. Internally, on the other hand, the search request transmitted by the search interface 3 can also be received, and the plurality of patent documents stored in the search platform 1 can be searched according to the search request, and the content of the patent file meeting the search requirements can be responded to the search. Interface 3. Each patent file retrieved by the retrieval platform 1 from the remote system 2 includes a plurality of attribute contents, including the abstract, the technical field to which the invention belongs, the prior art, the inventive content, the implementation manner, and the patent application scope, or It is a paragraph, a sentence, a word, and the like that further include at least one of the abstract, the technical field to which the invention belongs, the prior art, the inventive content, the embodiment, and the patent application. Further, the remote system 2 includes, but is not limited to, a US patent. The Patent Office (USPTO), the Japanese Patent Office (JPO), the Korean Patent Office (KIPRIS), and the European Patent Office (EPO) have multiple patent file storage databases, while the search interface 3 provides a search page for the user to enter at least one. Search condition.

The processing flow of the retrieval platform 1, the remote system 2 and the retrieval interface 3 includes the following steps:

(100) Drawing multiple patent documents.

(101) Parsing the plural attribute contents of the patent documents, respectively generating the same attribute content and different attribute contents.

(102) Transfer the same and different attribute contents to a plurality of storage locations.

(103) The same attribute content is stored in the same storage location, and the different attribute contents are respectively stored in different storage locations.

(104) Receiving a search request.

(105) The alignment search request is identical or partially identical to the attribute content in each storage location.

(106) generating a search result and responding the search result to the search interface 3.

In summary, the detailed description of the process flow can be further referred to the third and fourth figures, which are schematic diagrams of the search platform block diagram and the plural attribute content of a patent document according to a preferred embodiment of the present invention, which can be clearly seen from the figure. In this preferred example, the search platform 1 can include a data capture unit 11 and enable the data capture unit 11 to periodically retrieve the plurality of patent documents of the remote system 2, and the data capture unit 11 is connected. There is a data parsing unit 12, which can analyze and deconstruct the complex attribute content of each patent document, and generate complex and identical attribute contents, that is, the data parsing unit 12 can analyze and deconstruct the patent documents. And the abstract of each patent document, the field of the invention B, the description of the related art C, the summary of the invention D, the implementation ( Detailed description of the invention) E and the contents of the patent claim scope (what is claimed is) F are disassembled from each patent document, of course, The plural attribute content is defined to include paragraphs, sentences, and words including Abstract A, Technical Field B to which the invention belongs, Prior Art C, Invention Content D, Embodiment E, Patent Application Range F, and at least one of them, for example, Paragraphs F1, F2, sentences F11, F12 and words F111, F112 of the patent application scope F are similarly disassembled from the patent documents by the data analysis unit 12. Moreover, the data parsing unit 12 forms the same attribute content into a group. For example, the abstracts A of each patent document form the same group. In other words, the different attribute contents form different groups, for example, the abstract A of each patent document and The patent application scope F will form different groups, and the complex attribute contents in each group will be sequentially transmitted to the plurality of storage locations 13, after which the plurality of storage locations 13 will respectively perform storage in the same and different attribute contents. , the same attribute content is stored in the same storage location 13, for example, the abstract A of each patent document is stored in the same storage location 13, in other words, the different attribute content is stored in different storage locations 13, such as patent documents The abstract A and the patented range F are stored in different storage locations 13, respectively. The plurality of storage locations 13 may be defined as different server devices, or different storage spaces of the same server device, such as different hard disks located in the same server device, but not limited thereto. Further, the interconnection manner of the plurality of storage locations 13 may be a parallel architecture or a hierarchical architecture.

In the preferred embodiment, reference may be made to the fourth figure, and the fifth figure is a hierarchical architecture diagram of the plurality of storage locations of the present invention. The hierarchical architecture is used to describe that each storage location of the first level includes Abstract 13A, the technical field 13B to which the invention belongs, the prior art 13C, the invention content 13D, the embodiment 13E, and the patent application scope 13F, and the storage locations of the second hierarchy respectively include the abstract, the technical field to which the invention belongs, the prior art, and the inventive content. , the embodiment, and the paragraphs of the patent application scope, the third-level storage locations respectively include the abstract, the technical field to which the invention belongs, the prior art, the inventive content, the embodiment, and the sentences in the paragraphs of the patent application scope, the fourth-level storage location The words included in the abstract, the technical field to which the invention pertains, the prior art, the inventive content, the embodiment, and the sentences in the scope of the patent application are included.

For example, in the patent application scope F, the paragraphs F1, F2, sentences F11, F12 and the words F111, F112 of the patent application scope F are also included, and the storage location 13F of the first hierarchy can store the patent application scope F, The plural storage locations 13F1 and 13F2 of the second class store the paragraphs F1 and F2 in the scope of application for patents. The judgment method of each paragraph can use the full stop, and the patent application scope is stipulated in the patent application scope. Each paragraph represents the rights of the patent application scope, so the rights are stored in different storage locations 13F1, 13F2 of the second level, and the multiple storage locations 13F11, 13F12 of the third level are stored in the respective rights. For each sentence F11, F12, each sentence can be judged by using a semicolon or a comma, and each sentence is stored in a different storage location 13F11, 13F12 of the third hierarchy, and the plurality of storage locations 13F111, 13F112 of the fourth hierarchy are stored. Each word in each sentence F111, F112, each word can be judged by the character capacity, the words in each sentence are stored in different storage positions 13F1 of the fourth level 11, 13F112, however, the above description is only a preferred embodiment, but is not limited thereto.

Moreover, the parsing manner of the above-mentioned data parsing unit 12 includes, but is not limited to, an n-gram (n-gram). In other words, after receiving each patent document, the complex attribute content in each patent file is parsed by using an n-gram. Such a parsing program usually performs cutting analysis into a preset storage unit for each segment in the patent document, for example, when the storage unit is set to a word, and the patent application scope F in the patent document has "a heat dissipation structure of a lamp". For the sentence, the words to be parsed include "a kind", "species light", "lighting", "having", "scattering", "heating", "hot knot", "structure", etc. . These words are then stored in the word storage locations 13F111, 13F112 of the specified fourth level, respectively.

Further, the retrieval platform 1 of the preferred embodiment of the third embodiment further includes at least one logic operation unit 14, and at least one logic operation unit 14 is connected with a plurality of storage locations and a retrieval interface 3, respectively. The at least one logical operation unit 14 can be a microprocessor or a microcontroller (MCU), but is not limited thereto. The at least one logical operation unit 14 is configured to receive the search request transmitted by the search interface 3, and simultaneously read the attribute contents in the plurality of storage locations according to the search request, so that at least one logical operation unit 14 simultaneously performs the same with each storage location. Comparison program.

In summary, the hierarchical decentralized architecture disclosed in the present invention, when the retrieval request is transmitted through the search interface 3, each storage location separately performs the analysis comparison procedure of the retrieval request, so that the patent documents increase day by day. In this case, the retrieval processing speed of many patent documents can be accelerated.

In another preferred embodiment, the plurality of patent documents of the remote system 2 can also be pre-downloaded to a readable record medium, so that the readable record medium stores a plurality of patent documents, and the plurality of recordable media can be read. After the patent document is retrieved by the retrieval platform 1, a subsequent analysis procedure is performed. For example, the user may purchase a plurality of patent documents from the other party and store the patent documents in the search platform 1. At this time, the data capture unit 11 extracts the patent documents and causes at least one logical operation unit 14 to receive In these patent documents, at least one logical operation unit 14 then transmits each patent file to the data parsing unit 12 for the aforementioned parsing procedure. Thus, the source of the patent file to be stored or parsed by the search platform 1 may not be limited to the remote system 2 or the readable record medium storing the plurality of patent documents, or may be included in other forms or pre-stored from other servers. Select from the location.

Please refer to the sixth figure, which is a flowchart of the retrieval process according to the preferred embodiment of the present invention. It can be clearly seen from the figure that in the preferred example, the detailed processing flow of the retrieval platform 1 and the retrieval interface 3 includes The following steps:

(200) Provide a search interface 3 for receiving retrieval requests.

(201) Entering a search request into the search interface 3, wherein the search request includes at least one search condition.

(202) Determine whether the search request includes a plural search condition, and if so, proceed to step (203), and if no, proceed to step (205).

(203) coupling the complex search condition and generating a search parameter.

(204) Aligning the search parameters with the attribute contents of the plurality of storage locations, whether they are identical or partially identical, if yes, proceeding to step (206), and if not, proceeding to step (207).

(205) Whether the content of each attribute of the comparison search condition and the plurality of storage positions is identical or partially identical, and if so, proceeding to step (206), and if not, proceeding to step (207).

(206) Aligning the attributes that match the exact same or partially identical attributes.

(207) Generate a search result.

(208) Responding to the search result to the search interface 3.

In the above, the search program for the foregoing search process can be stored in a readable record medium, and when the search program is loaded into the device, and the search interface 3 is connected to the search platform 1 via a wireless or wired network, The user can input a search request through the search interface 3, and the search request is transmitted to at least one logical operation unit 14 of the search platform 1, and each logical operation unit 14 determines whether the search condition of the search request is a complex number. When the search condition is plural, the search conditions are coupled, and the search parameters are generated for subsequent comparison procedures. If the search condition is one, the search conditions are not coupled, and the comparison program is directly performed.

The search condition may be at least one of the preset keyword group, the search term input by the user, the corresponding attribute content, and the weight condition, and the plural keyword word is included in the preset keyword group. The keyword words are usually selected by a technical expert to select key words in the technical field, which is to initialize the preset data, and the user can add the search condition by selecting the method to increase the accuracy of the search in the technical field. . On the other hand, the user can also set the search condition by inputting the search term by himself, and the setting of the keyword word or the input of any word can further correspond to the attribute content, and even more, the above The search condition is coupled with a number of conditions, which can help define the keyword group and the search importance ranking of the user's arbitrarily input words, or the search of the user input word or preset keyword word and different corresponding attribute content. Priority sequence. For example, when the weight condition of the keyword word is (2) and the weight condition of any input word is (1), it means that the result of the keyword word after the comparison program is prioritized in any order. Before the word is entered, or the search condition search for the keyword term is prioritized, however, the interaction of the weight condition with the keyword group and any input words or the priority sequence definition can be designed by the user, and is not Limited.

Then, when the comparison program is performed, the search parameters generated by the at least one search condition may be used, or the search conditions may be compared with the attribute contents of the plurality of storage positions, and at least one logical operation unit 14 simultaneously reads and Search for the content of the attribute in the plural storage location that is identical or partially identical to the search parameter or search condition. If there is, the attribute content of the same or partially the same will be retrieved in sequence, however, regardless of whether the multiple storage location is The attribute content having the same or partially identical to the search parameter or the search condition is returned to the search interface 3 by at least one of the search results generated by the logical operation unit 14 so that the user can obtain the search interface 3. Know the search results.

The aforementioned comparison program is especially a program of data mining. This data mining program can be used but is not limited to an association rule, a part of speech analysis or an alignment algorithm, such as the Smith-Wattman algorithm (smith- Waterman) At least one of the above, that is, the algorithm or analysis method of each data mining can also be used in combination with each other.

Please refer to the seventh figure, which is a flowchart of the search process according to another preferred embodiment of the present invention. It can be clearly seen from the figure that this part is different from the preferred embodiment in another preferred embodiment. In the original architecture, an additional comparison program is added, that is, the processing flow of the retrieval platform 1 and the retrieval interface 3 includes two or more comparison programs in another preferred embodiment. Includes the following steps:

(300) Provide a search interface 3 for receiving retrieval requests.

(301) Entering a search request to the search interface 3, wherein the search request includes at least one search condition.

(302) Determine whether the search request includes a plural search condition, and if so, proceed to step (303), and if no, proceed to step (305).

(303) Coupling the complex retrieval conditions and generating retrieval parameters.

(304) Performing a first comparison procedure, comparing whether the search parameters and the attribute contents of the plurality of storage locations are identical or partially identical, and if so, performing step (306), and if not, performing step (307).

(305) Performing a first comparison procedure, comparing whether the content of each attribute of the search condition and the plurality of storage locations is identical or partially identical, and if so, performing step (306), and if not, performing step (307).

(306) A matching search result is obtained by matching the identical or partially identical attribute content.

(307) Performing a second comparison procedure to mine the correlation between the complex attribute content of the same or partially identical to the at least one retrieval condition and the attribute content of the plurality of storage locations.

(308) Generate a data mining result.

(309) Respond to the data mining results to the search interface 3.

Similarly, the retrieval processing flow of this other preferred embodiment is the same as the retrieval processing flow of the preferred embodiment of the sixth embodiment. The difference between the two is that, after the at least one logical operation unit 14 completes the comparison program, it is assumed that the search result after the comparison program is extracted to the attribute content that is identical or partially identical to at least one retrieval condition or retrieval parameter. After that, the second data comparison program will be performed. In detail, the first comparison program or the second comparison program can be used but is not limited to the association algorithm, word class/speech analysis method. Or at least one of the above algorithms, such as the Smith-Wattman algorithm, that is, the algorithm or analysis of each data mining can also be used in combination with each other. Wherein, regarding the association algorithm, the algorithm may be used to perform an accountability ratio correlation analysis of the attribute content of the attribute content that is identical or partially identical to the at least one search condition or the search parameter and the plurality of storage locations, for example, when The search conditions include any input words of "gastric ulcer" and attribute contents of "application patent scope", and the patent application scope corresponds to any input word, at least one logical operation unit 14 will couple the search conditions. After the comparison, a search result is generated, and the search result is to search for a plurality of identical or partially identical attribute contents in each storage location stored in the patent application scope. However, the storage locations of such comparisons naturally include the storage location of the first-level patent application scope, the storage location of the rights of the second-level patent application scope, and the claims of the third-level patent application scope. The storage location, the fourth-level application patent scope, the storage location of each word in each sentence. Then, at least one logic operation unit 14 further extracts, according to the search result of the search, whether the attribute content related to the search result is included in each storage location. The data mining algorithm includes but is not limited to the Apriori algorithm. Search for the search results of the same or partial "stomach ulcer" words in the scope of patent application, and further use the Apriori algorithm to obtain the "proton pump inhibitor" and "stomach ulcer" in each storage position. The probability of occurrence is higher than the preset value. This preset value can be set by the user. Thus, it can be known that the gastric ulcer is highly correlated with the proton pump inhibitor.

In another aspect, the data mining program may also use, but is not limited to, a part-of-speech/speech analysis method, and the analysis method is used to execute the attributes of the complex attribute content and the plural storage location that are identical or partially identical to the at least one search condition. Analysis of part of speech location relevance of content. For example, when the data mining program includes a word class/speech analysis method and an association algorithm, similarly, when the search condition includes any input word of "stomach ulcer" and the attribute content of "application patent scope", and Correlation algorithm concludes that gastric ulcer and proton pump inhibitors are highly correlated, and at the same time, by word/lexical analysis to determine the relationship between gastric ulcer and proton pump inhibitors as the main words and words, gastric ulcer can be further presumed. The drug solution is a proton pump inhibitor.

With regard to the Smith-Wattman algorithm, this algorithm is an alignment of the attribute contents of the complex storage location for the retrieval request, in other words, the comparison can be performed separately for the stored plural storage locations, and for each The comparison result of the storage location is defined by the penalty score calculation method, and the score calculation method includes, but is not limited to, a negative division into a positive score, wherein the negative score is defined as having no similar result after the comparison, and the positive score is The representation has similarity, and the scores are added to the degree of similarity, so that the higher scores represent higher degree of similarity. In order to simplify the calculation, the negative scores are also replaced by zero points, and may also be defined as positive scores, but are divided into fractions. size.

As another example, the above comparison method can facilitate the comparison of patent documents without defining clear technical content. In detail, the above comparison method can not only compare the content that is the same or similar to the search requirement, but also Assist in finding unclear but relevant technical content. For example, when the search request includes a general serial port (USB) search condition, and the search platform 1 stores a patent file, the disclosure has a "connection" The improved structure of the device, but not explicitly defined as the technology of such universal serialization, can be mined as a search result by using the association algorithm and the word/speech analysis method, that is, the "one connector improved structure" If the patent document reveals "signal pin and power pin", "4pin", and "5pin", you can use the correlation algorithm to find out the probability of occurrence, or combine the word/speech analysis to find out When "transmission" is a verb, whether the word "signal and power" has a relevant content. In this way, patent documents that do not explicitly reveal the structure of the universal serial port connector can be unearthed. However, this example is merely illustrative of the invention and is not limited to the file content of this type.

Referring to FIG. 8 again, it is a flowchart of a search process according to still another preferred embodiment of the present invention. It can be clearly seen from the figure that when the search request input in the program includes at least one file (Document), The definition of a document includes the content of a paragraph, a sentence, a word, and the like. For example, the document may be any equal aspect such as a patent document or a technical document. This part is different from the preferred embodiment in that, in another preferred embodiment, an additional file parsing program is added under the original architecture, that is, the present invention can further increase the file parsing program to form Another method of processing data retrieval, in other words, an analysis procedure for performing retrieval conditions before the comparison program, the retrieval processing flow includes the following steps:

(400) Provide a search interface 3 for receiving retrieval requests.

(401) Enter a search request into the search interface 3, wherein the search request is at least one file.

(402) Parsing the at least one technical file or patent file and generating a plurality of identical and different attribute contents.

(403) Determine whether the search request includes a plural search condition, and if so, proceed to step (404), and if no, proceed to step (406).

(404) Coupling the complex retrieval conditions and generating retrieval parameters.

(405) Aligning whether the search parameters and the attribute contents of the plurality of storage locations are identical or partially identical, and if so, proceeding to step (407), and if not, proceeding to step (408).

(406) Whether the content of each attribute of the comparison search condition and the plural storage location is identical or partially identical, and if so, proceeding to step (407), and if not, proceeding to step (408).

(407) A matching search result is obtained by matching the identical or partially identical attribute content.

(408) Generate a search result.

(409) Responding to the search result to the search interface 3.

In detail, when the search interface 3 receives the search request, the search interface 3 transmits the search request to the search platform 1, and at least one logical operation unit 14 of the search platform 1 determines whether the search condition is a file form, if the search condition When it is in the form of a file, at least one logical operation unit 14 transmits the file to the data parsing unit 12 for parsing, for example, the aforementioned n-gram. Of course, other parsing programs may be substituted instead of being limited thereto. After the data parsing unit 12 completes the parsing process and forms the complex attribute content, at least one logical operation unit 14 receives the attribute contents and performs a comparison program. This comparison program is described in detail by the Smith-Wattman algorithm, but is not limited thereto.

In detail, when the file is parsed into the content of the complex attribute, the algorithm can perform edge-wise comparison on the content of the attribute content and the content in the plural storage location respectively, in detail, if the attribute content of the retrieval is required It has the same prefix and suffix as the attribute content of the plural storage location, that is, when the same word is generated after the comparison, it is defined as the prefix, and then the search can be started in the same order. The same word is defined as the suffix. And the same prefix and suffix will form a segment, which may be a sentence or a paragraph. The same order of the segments has exactly the same or similar content, and the definition of the approximation can be set by the score calculation method, and the prefix and If the same order of the suffixes has exactly the same content, the comparison result of this retrieval request will be intercepted.

Therefore, such an algorithm must be in accordance with a certain string or word expression order in order to determine whether it is similar or identical to the content of the technical documents or patent documents in the search requirements. Further, since the plural attribute contents are pre-stored in different storage locations, subsequent retrieval procedures can be performed, and different retrieval conditions can be provided for comparison benchmarks, so that the comparisons will be identical or similar except for the comparison. The search result will also be defined according to the stored position, that is, if the comparison result is located at the paragraph position, it will indicate the same paragraph that appears in the document, for example, if the comparison result is in the scope of patent application, Mark in the scope of the patent application, or even which of the first few rights or rights.

It can be known from the above retrieval method that when the comparison has the same word and is defined as the prefix, the comparison will be performed from the beginning of the word, and the other parts will be ignored, and on the other hand, one can be further judged. In the attribute content, for example, the "patent application scope" has more similar or identical comparison results, and then compares the similar or similar attribute contents, which can save a lot of comparison time increase comparison efficiency. .

Furthermore, in still another embodiment, the probability of further determining the probability, that is, the probability that the retrieval condition is displayed in each storage location 13 of the retrieval platform 1 after a comparison condition is entered, is assumed to occur. If the patent document or the content of a certain attribute is significant, the patent document or the content of the attribute may be further compared, or the display result may be directly executed.

Of course, if the number of patent documents stored in the search platform 1 is more, the use of this probability judgment method saves the search time, for example, for the singular plural attribute content or for a patent document. The attribute content is further subjected to subsequent comparison, which can speed up the retrieval.

In this way, for a patent engineer who has written a patent specification for a long time, the technical file provided by the inventor can be input to find the relevant patent document content, so that the content of the prior art can be quickly obtained, and the patent specification can be quickly completed. Writing, improve writing efficiency. For the R&D engineer, it is also possible to quickly find the patents that have been applied for, based on the contents of the documents that have been published by others, so as to know the R&D direction of other companies or individuals. The field of development.

For example, when the patent engineer inputs a technical file to the search interface 3, the search interface 3 transmits the technical file to the retrieval platform 1, and at least one logical operation unit 14 of the retrieval platform 1 uses the data analysis unit 12 for the content of the technical file. Performing analytical deconstruction to parse the complex attribute content, such as at least one of a paragraph, a sentence, and a word, and then at least one logical operation unit 14 compares the parsed complex attribute content to determine each storage location 13 Whether there is data in the same or similar to the content of the plural attribute in order to generate the search result. In the comparison process, the attribute content of the retrieval request will respectively compare the contents in the storage location, and the comparison manner, when judging that there is no similar or identical content, will be separately marked or scored, and the judgment has Similar or identical content will be intercepted or scored in addition to being tagged or scored for transmission and display on search interface 3. Because the alignment program uses a segmented alignment method in conjunction with the distributed architecture of the present invention, the data is quickly retrieved and a search result is generated.

On the other hand, the Smith-Wattman algorithm and the association algorithm can define the keyword words in the search request, that is, the search results that are the same or similar to the search requirements are stored in the storage location and defined. This is a keyword word, which can be used as a convenience and speed for the user to search after the assistance, so that the user can set the subsequent search request according to the previous search result, so that the search analysis can be performed continuously.

The foregoing comparison method can be expressed in the comparison between the document and the file, that is, the retrieval requirement can include a technical document or a patent document, input the technical document or the patent document, and compare according to the content of the technical document or the patent document. .

In summary, in the architecture in which the search platform 1 has a plurality of storage locations, at least one logical operation unit 14 can perform data mining procedures on each storage location at the same time, thereby shortening the operation time of data mining.

Please refer to the ninth embodiment, which is a flowchart of a search process according to still another preferred embodiment of the present invention. It can be clearly seen from the figure that this part is different from the preferred embodiment in another preferred embodiment. An additional recombination word analysis program is added under the original architecture, that is, the search processing flow for the retrieval platform 1 and the retrieval interface 3 further includes a recombination word analysis program, and this further preferred embodiment is produced. Type, this retrieval process includes the following steps:

(500) Provide a search interface 3 for receiving retrieval requests.

(501) Entering a search request into the search interface 3, wherein the search request includes at least one search condition.

(502) Coupling to retrieve a plurality of combinations of at least one search condition and generating a word retrieval parameter.

(503) Whether the matching word search parameter and the attribute content of the plurality of storage locations are identical or partially identical, and if yes, proceeding to step (504), and if not, proceeding to step (505).

(504) Aligning the attributes that match the exact same or partially identical attributes.

(505) generate a search result

(506) Respond to the search result to the search interface 3.

In the above, when the retrieval platform 1 receives the retrieval request transmitted by the retrieval interface 3, at least one logical operation unit 14 of the retrieval platform 1 performs a recombinant word analysis program, and the manner of the program may include, but is not limited to, a recombinant word analysis method. That is, each logical operation unit 14 causes at least one search condition required by the search to generate a plurality of combined word search parameters, that is, when the search condition is an English block, each of the English blocks can be The letters are arranged in different English blocks, or when the search condition is English, each block of the English phrase can be arranged into different phrases. For example, the search conditions include "Tired nerves". The arbitrary input words and the attribute contents of the "implementation method", after the reorganized word analysis program, the search conditions may also include any input words of the "Tense driver" and the corresponding attribute contents of the "implementation method", so that it can be found Relevant meanings, so that the search conditions can cover a wide range of related words, and after generating the word retrieval parameters, "Tired nerves" and "Tense driver" can be used. By comparing the storage location of the "implementation" attribute content, the search result can be responded to the search interface 3, so that the diversity and broadness of the search can be achieved.

After the word correlation is linked, the related technology can also be stored in the device through learning, so that the retrieval speed can be accelerated.

In summary, the search results in the foregoing preferred embodiments are displayed on the screen of the search interface 3, which may be in the form of a diagram or a list, or the manner of the diagram or the list may be searched by the search interface 3. The received search conditions are affected. In detail, when the search condition received by the search interface 3 is performed, the foregoing comparison process is performed, and then the generated search result icon or the column arrangement of the table column includes High similarity ranks first, the similarity is low, and even the default condition value received by the search interface 3 (for example, according to the application date, public date or some weight setting, etc.), such as the search result The degree of similarity in the patent application range is higher, or the higher frequency of the preset keywords is ranked first, and the above plural setting methods can also be coupled to each other. Of course, the ordering of these columns is It can be set differently depending on the usage habits and is not limited to the above description.

In other respects, a further preferred embodiment of the search result icon or the manner of the list, please refer to the tenth figure, which is a schematic diagram of the search result according to the preferred embodiment of the present invention, which can be clearly seen from the figure, if the search result The response to the search interface 3 can be a tree structure diagram of relevance. In detail, if the input search condition is "associated test value" file and "patent application scope", the comparison program will be performed, and Corresponding to the two search conditions, when the search result is generated, in addition to the associated patent documents including the "associated test value", the associated words or the main words, verbs, The words are retrieved together, and each associated word can be linked to the "associated test value" in a tree structure. For example, a patent application scope can be retrieved with the associated test value. Patent document, after the publication of this patent document, will list the rights in the first layer of the tree structure, and the related subject words or the words "device" and "test value" listed in the tree Knot The second layer of the structure, and each word list will also display its corresponding weight value.

In addition, another example is shown in FIG. 11 , which is a schematic diagram of a search result according to another preferred embodiment of the present invention. It can be clearly seen from the figure that if the search result is returned to the search interface 3 For a tree structure diagram, the representation of the tree structure may be based on the respective paragraphs, sentences and words of the abstract, the technical field, the prior art, the inventive content, the embodiment or the patent application scope, and A tree arrangement is formed according to a predetermined priority sequence. In detail, similarly, the input search condition is a file having the "associated test value" and the "application patent scope", and the two search conditions are correspondingly matched, and when the search result is generated, a patent application can be retrieved. A patent document with the scope of the associated test value. If the scope of the patent application is expanded, a tree structure can be formed. The tree structure of the first layer can be a preset device or method, and is noted as Open, semi-open or closed, expressed as the main right, and may further note whether there is "further inclusion", which is marked as a subsidiary right, and the second level is indicated in each major right or subsidiary An associated word that defines a subject or a subject in a right, such as a "weighted value" or "invalid" word.

Further, the search platform 1 may be updated according to the number of patent documents or content stored in the remote system 2, and the update manner may be determined by the data capturing unit 11 to determine whether the remote system 2 has at least one patent to be updated. And when the data is judged to be updated, the data retrieval unit 11 sequentially retrieves the patent documents that need to be updated from the remote system, and after receiving the data, the data analysis unit 12 receives the updates. The patent documents are analyzed and deconstructed for the content of the plural attributes in the updated patent documents, and the identical and different attribute contents generated after the analysis of the deconstruction are separately transmitted to the plurality of storage locations 13.

The processing or retrieval program for any of the foregoing processing or retrieval processes can be stored in a readable recording medium, and can be executed after a device that can read the program loads the retrieval program. The device that can read the search program can be a computer, a device capable of executing a thread, or a device for executing a single thread, and the recordable medium can be a network server. The search program is installed in the device via online download, or stored on a disc, but is not limited thereto.

However, the above description is only an example of the implementation of the present disclosure, and the scope of the present invention cannot be limited thereto. All changes and modifications made to the scope of the present invention should remain within the scope of the present invention.

1. . . Search platform

11. . . Data acquisition unit

12. . . Data analysis unit

13. . . Storage location

14. . . Logical unit

2. . . Remote system

3. . . Search interface

A. . . Summary

B. . . Technical field to which the invention belongs

C. . . Prior art

D. . . Summary of the invention

E. . . Implementation

F. . . Patent application scope

F1. . . Patent application passage

F2. . . Patent application passage

F11. . . Patent application scope sentence

F12. . . Patent application scope sentence

F111. . . Patent application term

F112. . . Patent application term

13A. . . Summary storage location

13B. . . Technical storage area to which the invention belongs

13C. . . Prior art storage location

13D. . . SUMMARY OF THE INVENTION Storage location

13E‧‧‧Implement storage location

13F‧‧‧Requested patent range storage location

13A1‧‧‧ Summary paragraph storage location

13B1‧‧‧Technical field paragraph storage location

13C1‧‧‧Previous technical paragraph storage location

13D1‧‧‧Inventory paragraph storage location

13E1‧‧‧Implementation paragraph storage location

13F1‧‧‧Application patent range paragraph storage location

13A11‧‧‧Abstract sentence storage location

13B11‧‧‧Technical field sentence storage location

13C11‧‧‧Previous technical sentence storage location

13D11‧‧‧Inventive sentence storage location

13E11‧‧‧ Implementation sentence storage location

13F11‧‧‧Application for patent range sentence storage location

13A111‧‧‧Abstract word storage location

13B111‧‧‧Technical field storage location

13C111‧‧‧Previous technical word storage location

13D111‧‧‧Invented content word storage location

13E111‧‧‧ Implementation word storage location

13F111‧‧‧Requested patent range word storage location

The first figure is a process flow diagram of a preferred embodiment of the present invention.

The second figure is a schematic diagram of a platform architecture of a preferred embodiment of the present invention.

The third figure is a block diagram of a search platform of a preferred embodiment of the present invention.

The fourth figure is a schematic diagram of the attribute content of the patent specification of the present invention.

Figure 5 is a schematic diagram of a hierarchical architecture of a plurality of storage locations in accordance with a preferred embodiment of the present invention.

Figure 6 is a flow chart showing the retrieval process of the preferred embodiment of the present invention.

Figure 7 is a flow chart showing the retrieval process of another preferred embodiment of the present invention.

The eighth figure is a flowchart of the retrieval process of still another preferred embodiment of the present invention.

Figure 9 is a flow chart showing the retrieval process of still another preferred embodiment of the present invention.

The tenth figure is a schematic diagram of the search results of the preferred embodiment of the present invention.

Figure 11 is a schematic diagram showing the results of a search for another preferred embodiment of the present invention.

Claims (18)

A processing method of a patent document retrieval platform of a cloud architecture, which can perform a retrieval process of a plurality of patent documents by using a retrieval platform, and the patent documents include plural attribute contents, and the steps thereof include the following steps: extracting a plurality of patent documents; The plural attribute content of the patent documents respectively generates the same attribute content and the different attribute content; the same and different attribute contents are transmitted to the plural storage locations; the same attribute contents are respectively stored in the same storage location, The different attribute contents are respectively stored in different storage locations; receiving the search request, when receiving the search request, the at least one search condition may be generated by the recombinant word analysis method to generate a plurality of combined word search parameters, and then The plurality of combined complex digital word retrieval parameters are compared with each attribute content of the plurality of storage locations; the comparison retrieval request is identical or partially identical to the attribute content in each storage location; generating a search result and The search results are echoed to the search interface. The processing method of the patent document retrieval platform of the cloud architecture described in claim 1, wherein the attribute content in the plurality of storage locations is an updateable program, and the data retrieval unit can determine whether a remote system is There is at least one patent document that needs to be updated, and when it is judged that the update is needed, the data retrieval unit then sequentially retrieves the patent documents that need to be updated from the remote system. And after extracting, using a data parsing unit to receive the updated patent documents, and analyzing and deconstructing the complex attribute contents in the updated patent documents, and analyzing the identical and different attributes generated after deconstruction. The content is then transferred to multiple storage locations. For example, in the processing method of the patent document retrieval platform of the cloud architecture described in claim 1, wherein the retrieval result may be a tree structure in response to the retrieval interface, and the representation manner of the tree structure may be summarized and invented. The respective paragraphs, sentences and words of the technical field, prior art, inventive content, embodiment or patent application range respectively capture part of the content and are arranged in a tree according to a predetermined priority sequence. The processing method of the patent document retrieval platform of the cloud architecture described in claim 1, wherein when the retrieval request is input to the retrieval interface, the retrieval request includes at least one retrieval condition, and at least one retrieval condition includes At least one of the keyword group, the weight setting condition, the search content input by the user, and the corresponding attribute content. A search method of a patent document retrieval platform of a cloud architecture, the retrieval method is a retrieval program, and is stored in a readable record medium. When a device capable of reading the retrieval program is loaded into the retrieval program, the following execution is performed. Step: providing a search interface to be connected to a search platform, wherein the search interface can receive the search request, and have a plurality of storage locations in the search platform, and each storage location is stored in the same storage location for the preset identical attribute content, The different attribute contents are stored in different storage locations; Entering the search request to the search interface, and the search request includes at least one search condition; transmitting the search request to the search platform; comparing the search request with the attribute content of the plurality of storage locations in the search platform; capturing exactly the same or partially the same as the search The required attribute content; and generate a search result and respond to the search result to the search interface. For example, the patent document retrieval platform of the cloud architecture described in claim 5, wherein when the attribute content that is identical or partially identical to the retrieval requirement is retrieved, a data mining program may be performed, and the data mining program includes a word class. / part of speech analysis. For example, the patent document retrieval platform of the cloud architecture described in claim 5, wherein the data mining program can be performed after extracting the attribute content that is identical or partially identical to the retrieval requirement, and the data mining program includes Smith. - Waterman algorithm. A search method of a patent document retrieval platform of a cloud architecture, the retrieval method is a retrieval program, and is stored in a readable record medium. When a device capable of reading the retrieval program is loaded into the retrieval program, the following execution is performed. Step: providing a search interface to be connected to a search platform, wherein the search interface can receive the search request, and have a plurality of storage locations in the search platform, and each storage location is stored in the same storage location for the preset identical attribute content, The different attribute contents are stored in different storage locations; the search request is input to the search interface, and the search request includes at least one search bar Transfer the search request to the search platform; compare the search requirements and the attribute content in the plural storage location; retrieve the attribute content that is identical or partially identical to the search request; execute the data mining program; and generate a search result and respond Search results to the search interface. The method for retrieving a patent document retrieval platform of the cloud architecture as described in claim 8 wherein the data mining program executed includes a Smith-Wattman algorithm. For example, the method for searching a patent document retrieval platform of the cloud architecture described in claim 8 of the patent application, wherein the data mining program executed includes a part-of-speech/partition analysis method. The method for searching a patent document retrieval platform of the cloud architecture described in claim 5 or 8, wherein the method for storing the plurality of attribute contents is to retrieve a plurality of patent documents from a remote system through a data retrieval unit. Then, the data acquisition unit transmits the patent documents to a data analysis unit, so that the data analysis unit receives the patent documents, and then performs an analysis deconstruction program of the plurality of attribute contents of the patent documents, and the analysis destructor program generates the The same and different attribute contents of the patent documents are respectively transmitted to the plurality of storage locations, so that the plurality of storage locations respectively receive the same and different complex attribute contents generated by the data parsing unit, and store the same attribute content in the same storage location. , and different attribute content is stored in different storage locations Set. The method for searching a patent document retrieval platform of the cloud architecture described in claim 5 or 8, wherein the attribute content in the plurality of storage locations is an updateable program, and the data retrieval unit can determine a remote end through a data retrieval unit. Whether the system has at least one patent file that needs to be updated, and when it is judged that the update is needed, the data acquisition unit sequentially retrieves each patent document to be updated from the remote system, and uses a data after the retrieval. The parsing unit receives the updated patent documents, and analyzes and deconstructs the complex attribute contents in the updated patent files, and separately transmits the identical and different attribute contents generated after the deconstruction analysis to the plurality of storage locations. The method for searching a patent document retrieval platform of the cloud architecture described in claim 5 or 8, wherein the retrieval result may be a tree structure in response to the retrieval interface, and the representation of the tree structure may be summarized The respective paragraphs, sentences and words of the technical field, the prior art, the inventive content, the embodiment and the patent application scope of the invention respectively extract part of the content and form a tree arrangement according to a predetermined priority sequence. The method for searching a patent document retrieval platform of a cloud architecture according to claim 5 or 8, wherein when the at least one retrieval condition is received to the logical operation unit, the at least one retrieval condition may be generated by using a recombinant word analysis method. A plurality of combined word retrieval parameters, and the plurality of combined complex digital word retrieval parameters are compared with the respective attribute contents of the plurality of storage locations. The method for searching a patent document retrieval platform of the cloud architecture described in claim 5 or 8, wherein if the retrieval request includes a plurality of retrieval conditions, the coupling process is performed, and a retrieval parameter is generated, and then the retrieval is performed. A comparison program of parameters and attribute contents in a plurality of storage locations. A patent document retrieval platform of a cloud architecture, the platform is capable of receiving at least one retrieval requirement, and performing a retrieval operation of a plurality of patent documents according to the retrieval requirement, and each patent document includes a plurality of attribute contents, and the platform includes: The data acquisition unit may retrieve a plurality of patent documents; a data analysis unit is connected to the data extraction unit, and the patent documents may be received, and the complex attribute content in the patent documents is analyzed and deconstructed, and the analysis is deconstructed. The generated identical and different attribute contents are respectively transmitted to the plurality of storage locations; the plurality of storage locations are connected to the data parsing unit, and respectively receive the same and different complex attribute contents generated by the data parsing unit, so that the same attribute content is obtained. Correspondingly stored in the same storage location, the different attribute contents are respectively stored in different storage locations; and at least one logical operation unit is correspondingly connected to the plurality of storage locations for receiving and computing the retrieval request and each storage location. Attribute content, generate search results. For example, the patent document retrieval platform of the cloud architecture described in claim 16 wherein the data acquisition unit further actively detects a remote system. Is there at least one patent document that needs to be updated, and the patent documents that need to be updated are sequentially taken. The patent document retrieval platform of the cloud architecture described in claim 16 wherein the plurality of storage locations can be defined as different storage locations of different server devices or a single server device.