Patent information analysis method is based on bibliometrics, with the help of other disciplines' knowledge and related tools. In the past, patent information analysis mainly extracted a large amount of patent information from patent documents manually, and analyzed and processed it by using relevant statistical methods combined with industry experience, so as to explore the information hidden behind patent documents and serve the decision-making of enterprise technology innovation management. So the main analysis methods at that time were original text analysis, simple statistical analysis, chart method based on simple statistics, dynamic vector method and so on.

1. Original analysis.

By searching competitors' patent specifications, carefully reading and analyzing, we can master the development characteristics of competitors' new products and technologies, including finding gaps, technical improvement, technical synthesis, patent technology principles, etc.

2. Simple statistical analysis.

According to the number of patent inventors, the number of patent applicants, the number of patent classification and the number of patent documents, statistical analysis was carried out respectively. Through the statistical analysis of the relevant situation, we can understand the current situation of scientific and technological progress in various countries, the transfer of technical research interests or hot spots, find out the concerns of current technical inventors and the development direction of this technical field to a certain extent, see the competition in a certain technical field, and even judge the most active field.

3. Statistical analysis of collocation.

Through the combination statistics of patent classification number, patentee, patent application date (authorized publication date) and patent application country in patent statistics, various statistical information is obtained, and then these statistical information is analyzed.

4. Keyword frequency statistics.

(1) Delete duplicate patent applications, and then extract some keywords with technical experiment concepts from patent items, abstracts and titles; (2) Statistical frequency of keywords; (3) Reading meaning is a logical combination of keywords with high probability of occurrence, and it is a re-understanding of technical concepts.

5. Statistics will be made after technical subdivision.

According to the principle of hierarchical tree, a technology is subdivided and its subordinate concepts are counted item by item.

6 index change chart and technical trends and characteristics comparison table.

The table of technical trends and characteristics is mainly used to reflect the technical trends and characteristics of different enterprises applying for patents in different years from the perspectives of technical fields and certain functions of products, so as to compare the technical development trends and directions of various enterprises. The main forms are: comparison of enterprise technology development in different years and technical fields, comparison of different scientific research topics, comparison of different scientific research topics in different enterprises, and regression analysis of various factors.

7. Vector dynamic model method.

Patent documents not only reflect the quantitative relationship of science and technology, but also imply the development direction of science and technology. So it is represented by the concept of vector. The application of vector model method is vector modeling of statistical dynamic data, and then evaluating and predicting the scientific development trend.

8. Patent citation analysis method.

This paper analyzes the phenomenon of citing references in patent literature, reveals its quantitative characteristics and memory rules, and evaluates the technical development trend accordingly.

9. Thematic data analysis method.

The so-called "thematic data analysis method" is to sort out, combine and analyze the geographical distribution and research content of a thematic document according to the dispersion degree of patent documents in the international invention classification table, and predict the countries with the most active innovation activities and key research fields in the world.

Second, the latest development of patent information analysis methods

With the popularization of computers, the development of information technology and network technology, patent information analysis has gradually shifted from manual processing to the era of using computers as tools. This provides great convenience for patent information analysis and promotes the development of patent information analysis methods to automation, intelligence and visualization.

1. Computerized quantization.

In order to quantitatively analyze hundreds of patent documents contained in each field, it is necessary to search, screen, count and draw charts on the description items such as application date, main classification number, priority applicant country, applicant and publication number of these patent documents. It is obviously time-consuming and laborious to use manual methods. Therefore, it is necessary to establish a patent information analysis database and use computer analysis and statistical methods to quantitatively analyze the patent documents in the database.

In the patent information analysis database, using the powerful function of EXCEL spreadsheet, the description items are sorted, and the patent application numbers of each year, country and company are screened out, and then the above items are quantitatively counted, so as to make a targeted statistical chart and understand the patent development in this field from all aspects. Specifically, it mainly includes: ① the time distribution of the number of patent applications can measure the future development trend of a certain technical field; (2) Classification of technical topics in patent documents, which can classify technical topics and make a distribution map of technical topics; (3) By counting the number of domestic patent applications in priority countries, we can know the technical strength of each country in this field; (4) Counting the number of patents owned by each company can reveal the technical and economic strength of each company in this technical field; ⑤ The company's foreign patent applications, count the number of patent applications in more than two countries, and draw the distribution map of the company's foreign patent applications; ⑥ The distribution of the number of patent applications in different countries, and finding out the competitors faced in different countries can help us understand the degree of market competition of foreign companies in these countries; ⑦ Statistical analysis of patent applications in various countries can help us understand the competition of companies in various countries.

2. Patent information analysis method based on similarity function.

The similarity function mentioned here includes: similarity function based on patent citation and similarity function based on term concurrency, and similarity function based on term concurrency in literature is also effective. Reliable data sources provide consistent indexing terms for each document, which are used to establish similarity functions based on term concurrency.

This function is easy to execute using SQL query statements. After calculating the similarity between documents, we use forced direct placement or self-organizing neural network technology to map documents into two-dimensional space to form document clusters to explore the relationship between documents. The force-oriented collocation method is that at the beginning, all documents are placed in the center of the plane, and the force between documents is inversely proportional to their distance and directly proportional to their similarity value; If the similarity value is greater than 0, it means attraction, otherwise it means repulsion; The direction of force in mechanics obeys Coulomb's law. Repeat this process until a stable document structure diagram is formed. Self-organizing neural network technology is to train rectangular self-organizing neural network by inputting variables with similar matrix behavior. Generally, a new self-organizing neuron matrix is adopted. After training, the density distribution of neuron connection weights in n-dimensional space matches the density distribution of training vectors.

Generally speaking, patent analysis can be divided into quantitative analysis and qualitative analysis.

1, quantitative analysis

Quantitative analysis is mainly based on the statistical analysis of the appearance characteristics of patent documents, that is, identifying relevant documents through the inherent indexing items of patent documents, then making statistics on relevant indicators, and finally explaining the changes of relevant data in different ways in order to obtain the information of dynamic development trend.

(1) Statistical object and angle

① Statistical objects are generally based on the number of patents. ② Statistics can be made from different angles according to patent classification, patentee, year and country. When the patent information is classified and counted, according to the number of patents in various fields, we can know which technical fields are active in invention, which technologies will be broken through and which technologies will be eliminated soon. If the patent information is counted according to the country, we can find the scientific and technological development strategy of the counted country and its position in various fields. This statistical result is helpful for people to understand the key points of scientific research development in various countries in a certain period. If we count the patents according to the patentee, we can find the important technology owners in a certain field, or which company has an important position in this field. (2) the main indicators of statistics

① Number of patents. The number of patents in a certain technical category can be used to measure the level of technical activities in this technical field; The number of patents applied by a company or patentee over the years reflects the occurrence, development process and development trend of its technological activities. The number of patents can be used to compare the output of technological activities and the intention of seeking industrial property protection in different countries in different periods and fields. ② The number of patents in the same family. The number of patent families of an invention reflects the breadth of the company's patent application field and the potential value of this invention. Because of the cost of translation and special legal help, it is much more expensive to apply for a patent abroad than in your own country. Only those inventions recognized by the company as having the most commercial value will be patented in many countries to protect the exclusive rights of future investment and product export. ③ Number of patent citations. The number of times a patent is cited by subsequent patents can reflect the importance of this patent, because an important patent will be accompanied by a large number of improved patents, and this important patent will be repeatedly cited by improved patents. Citation reveals the relationship between patents, which can be used to track the patent network corresponding to different technologies and find the patents at the intersection of different technologies. Unfortunately, the patent database in China can't provide the number of patents cited, which is a pity. (4) patent growth rate. The patent growth rate measures the percentage of the number of patents growing with time, which can show whether the change of technological innovation with time is increasing or slowing down. For example, the quarterly growth rate of patents is to compare the number of patents obtained by enterprises in a certain quarter with the number of patents obtained in the previous quarter, and calculate the percentage rate of increase or decrease of the number of patents obtained in this quarter compared with the previous quarter. The annual growth rate of patents is the percentage of patent growth compared with the previous year, which is used to measure the change of technical activities in the past year. ⑤ Scientific relevance. Scientific relevance measures the number of scientific papers or research reports cited by patents (subject patents) and the relationship between patented technology and cutting-edge scientific research. The value of scientific correlation is industry-dependent: the average value of scientific correlation in machinery industry is close to zero, while that in high-tech biochemical industry may be as high as 15. ⑥ Technology life cycle. The technology life cycle measures the average number of years of patent technology cited in the title page of enterprise patent application documents. Therefore, the technology life cycle can be understood as a period between the latest patent and the earliest patent. If the technology life cycle is short, it means that efforts are being made to develop a relatively new technology, and the development and innovation of this technology are very fast. The technology life cycle is industry-dependent, and the technology cycle of popular industries is relatively short, such as electronics for about 3-4 years, medicine for 8-9 years, shipbuilding 15 years. ⑦ Patent efficiency. Patent efficiency measures the patent quantity output created by a certain R&D expenditure, which is used to evaluate the scientific research ability and cost efficiency of the enterprise's patent quantity output within a predetermined time. The more patents are produced, the higher the patent efficiency, and the stronger the technological research and development capability of enterprises. (8) Patent implementation rate. Whether the patent implementation rate can be effectively implemented and whether it can bring about scientific and technological innovation is still unknown for those patented technologies that have not yet been implemented. The implementation of general invention patents goes through a development process, and development is not always successful. Many invention patent technologies have to be abandoned halfway or finally given up because of technical difficulties that cannot be solved or expected results can not be achieved under the existing technical conditions. The implementation of invention patent can be measured from the aspects of technical performance, economic benefits, social benefits, market factors, industrial development and production capacity, macro-environment and industrialization risks. The higher the patent implementation rate, the greater the contribution of patents to technological development and innovation, and the closer the combination with technological development. The patent implementation rate in China is only about 30%, which is far lower than that in Europe, America and Japan. Pet-name ruby industry standardization index. In the cross-industry horizontal comparison, the differences between industries bring trouble to the comparison of patent index values between different industries, so it is necessary to use industry standardized indicators. The industry standardization index value is obtained by dividing the index value of an enterprise by the average index value of the industry in which the enterprise is located. For example, there are 30 enterprises in the chemical industry, and the average value of their scientific relevance is 3.7, so the standard index value of each chemical enterprise's scientific relevance should be the index value of each enterprise's scientific relevance divided by 3.7. In this way, we can eliminate the different influences brought by different industries, and then find out the best-performing enterprises in each industry.

(3) The main contents of statistics

① Study on the time distribution of patented technology. That is to say, with time as the horizontal axis and the number of patent applications (or approvals) as the vertical axis, the trend is generally predicted by counting the changes of the number of patents with time. Any technology has a process of emergence, development, maturity and aging. The change of patent application and patent citation over the years can determine the development trend and active period of this technology, and provide basis for major decisions such as scientific research projects and technology development. By comparing the time distribution of patents in different technical fields, we can determine which technical fields are active and which are stagnant in a certain period of time. ② Study on the spatial distribution of patented technology. That is, by comparing the number of patents of different companies and enterprises, we can reflect their technical level and strength. Spatial distribution is generally used to identify competitors and analyze their technical strategies. By counting the patent applications of a certain technology category according to the patentee, we can get the distribution of a certain technology in different companies or enterprises, and know which companies or enterprises have invested more in this field, the patent activities are more active and the technical level is leading; By counting the patent frequency of the company in different technology categories, we can know the most active field of the company, that is, the key field of the company's development. In addition, by searching the same family patent of a patent, we can get the geographical distribution of these patent applications, so as to judge their commercial value and understand the key areas of a company's technology output; It can also provide a basis for technology introduction and information for products to avoid each other's protected areas.

(4) Measurement of statistical parameters in different stages of technical development.

Technology growth rate v: where a: the number of invention patent applications (or approvals) in that year; A: The cumulative number of applications for invention patents (or the cumulative number of approvals) goes back to 5 years. After several years of continuous calculation, the value of v is increasing, indicating that the technology is in the bud or growth stage. Technology maturity coefficient α:. Where a is the same as above, and b is the number of applications (or approvals) for utility model patents in that year. After several years of continuous calculation, the α value has decreased, reflecting the maturity of the technology. Technical aging coefficient β:. Where A and B are the same as above, and C is the number of applications (or approvals) for design patents or trademarks in that year. After several years of continuous calculation, the β value is getting bigger and bigger, which shows that this technology is becoming obsolete. New technology characteristic coefficient n: n = υ 2+α 2. Where υ is the technology growth rate and α is the technology maturity coefficient. It is a comprehensive index reflecting the emerging or aging of a technology. The greater the value of n, the stronger the characteristics of the new technology, indicating that it has more development potential. 2. Qualitative analysis

Qualitative analysis, also known as technical analysis, is to identify patents according to their technical content or "quality", and merge related patents according to their technical characteristics to make them orderly. This is quite different from the statistical analysis that only relies on the appearance characteristics of patent documents. Qualitative analysis is usually used to obtain information about technology trends, enterprise trends and specific rights. We can consider the contents of important patents from five aspects: the purpose, principle, material, structure and method of invention, and classify important patents according to their similarities and differences. If the patent content is based on principle, it means that this technology is not mature; If the patent content focuses on the diversity of uses, it shows that the technology has been put into practical use. In addition, according to the patent content list, the patents of major companies in a certain technical field can be analyzed, and the technical characteristics and development priorities of each company can be seen. According to the similarities and differences of technical contents, the related patents are divided into patent groups, and the changes of different patent groups owned by a company or patent groups in different periods are analyzed, so as to analyze and predict the key problems in the development process of a technology or product, the future development trend and application trend, and the relationship with other technologies.

Because it involves the specific content of technology, the work of qualitative analysis is heavy and complicated. Whether to use quantitative analysis or qualitative analysis depends on the problems to be solved and the patent data to be mastered. In fact, it is often necessary to combine qualitative analysis with quantitative analysis to achieve good results. For example, we can first determine which companies have technical advantages in a certain technical field through quantitative analysis (the number of patent applications or approvals can reflect the level of technical activities), and identify important patents in this technical field (the number of patents cited by subsequent patents reflects the importance of patents), and then conduct qualitative analysis on the important patents of these companies.

The quantitative analysis and qualitative analysis of patent information reflect the development status and trend of technology through quantitative change and internal qualitative change. There are both differences and inevitable connections between them. The classification of quantity needs to be based on quality, and the embodiment of quality needs to be through quantity. Therefore, in practical work, the combination of the two will achieve better results.