1, the continuation of the traditional role
It is pointed out in the IPC user's guide that IPC is a tool to unify the classification of patent documents in various countries. With the help of this international unified patent document classification system, an effective patent document retrieval tool can be established for patent offices and other users, so as to evaluate the novelty and creativity of patent applications.
In terms of information services, IPC also has the following functions:
(a) Using the classification table to arrange patent documents can make users easily obtain technical and legal information from them;
(b) As the basis for user-selective reporting of patent information;
(c) As a basis for investigating the existing technical level in a certain technical field;
(d) As the basis of industrial property statistics, so as to evaluate technological development in various fields.
In addition, IPC is also a tool for examination, management and statistics of patent offices.
2. As a tool to realize computer intelligent classification.
Needless to say, IPC is a product based on the historical condition that paper patent documents and information technology are not developed enough. Some principles and principles in its development seem a bit outdated from today's perspective. But it depends on the essence. The essence of IPC is a classified navigation index system, and the application of this system is to index patent documents. In today's highly developed information technology, an index system that has been tested by practice still has its indispensable practical significance for the application of information technology.
For intelligent classification, the basic solution is to convert IPC into machine-readable "electronic version". This machine-readable electronic version is actually an electronic language expression table with IPC as a tree structure framework. It not only transforms the current description methods of IPC technical terms and natural language or phrases into the description methods of technical terms and relational terms, but also the combination and logical relationship between words. Through computer programming, technical terms (including synonyms) and relational terms (including verbs and prepositions, etc.) are combined. In the eight parts of the coding application manual, statistics are made according to word frequency, position and relationship, and the relationship between words and the expression of logical relationship are analyzed and judged. Through weighted statistics and garbage disposal, the classification position is finally listed according to the degree of association with IPC electronic version. The listed categories can be several or more than a dozen. Because, at this time, we have not simply regarded them as the primary classification and secondary classification in the traditional sense, but as the index of application cases, thus creating conditions for the intelligent retrieval of patent documents in the next step.
As mentioned above, this reform will be a structural reform of IPC, which is very difficult and has a considerable workload, but once it is completed, it will be of great significance. At that time, there will be no need for special classification examiners to complete the manual classification of applications in the patent examination process. What needs to be done will mainly focus on the improvement and enrichment of the electronic watch by collecting feedback from outside the process, and this electronic language expression table will be maintained by many classification experts. The handling of confidential disputes is still in the examination department, and only the results of dispute handling should be summarized into the hands of experts. This work may have a long way to go, but the author believes that intelligent classification of patent applications is most likely to be the forerunner to realize intelligent patent examination.
3. It can be used as an important tool to realize computer intelligent retrieval.
As another field to be developed by IPC, it is to realize computer intelligent retrieval with the help of IPC, including intelligent determination or expansion of retrieval fields, intelligent analysis, selection of retrieval records and so on.
Computer intelligent retrieval is one of the goals pursued by the information technology industry at present, and it is also the direction for our bureau to improve the patent retrieval system in China. Realizing the intelligent retrieval of Chinese patent documents can fundamentally solve the problems of recall and precision, thus improving the efficiency and quality of examination. To solve the problem of intelligent retrieval, it seems that there is not much problem from the analysis of software and hardware conditions. At present, some commercial retrieval systems have initially realized intelligent retrieval based on thesaurus, but they are often vague and not suitable for patent literature retrieval because of their inherent shortcomings. The author thinks that the reason why the intelligent retrieval of patent documents stays on paper is mainly to determine the technical scheme to realize intelligent retrieval and a lot of basic work to be done to realize this technical scheme.
From the technical solution, simple full-text free word retrieval and simple manual indexing or automatic word segmentation indexing can not completely solve the patent literature retrieval problem. Because Chinese takes the words formed by the combination of words as the unit of text description, it causes more uncertainty. Searching for common words may not be a prominent problem. However, it is not enough to retrieve all-encompassing patent documents. Using IPC to search can make up for the above shortcomings. Because IPC classifies technical topics with the same attributes into the same category number according to scientific classification methods. Under the same classification number, we can not only find the same theme as the invention, but also find similar technical solutions. In addition, although many databases handle the source data well, such as depth indexing, their indexing has certain principles. For words that are too common, they are generally not indexed. If the patent classification is used as the index, it can not only contain these words, but also cover the technical characteristics, technical themes and even technical solutions of patent documents from the technical connotation. Therefore, it is necessary to adopt the retrieval mode of combining word, word and classified indexing to solve the problem of Chinese intelligent retrieval (manual indexing can be retained, but it should be used as an auxiliary means to deeply index some special fields and improve the source data).
To this end, we can play the role of IPC in two aspects. First, in addition to full-text retrieval technology, a set of thesaurus, including thesaurus, must be established for word retrieval. IPC and its supported keyword index can be the basic framework tool for generating the thesaurus. To some extent, the Patent Office has the conditions to become the basic production base of thesaurus in the field of science and technology in China. This thesaurus may have the following characteristics: compared with the thesaurus used in every industry, it may not be the best or the most popular, but it is not better or more popular than all industries. Secondly, the above-mentioned electronic language version IPC and classified indexing are important means to enrich retrieval, and will be applied to intelligent retrieval of patent documents as a realistic way. Using the domain division, classification system and location relationship of IPC can help us to carry out intelligent retrieval. For example, if we only input a main category or a technical field name or even a keyword into the retrieval formula, the intelligent retrieval system can automatically complete the retrieval within the relevant scope and get the retrieval requirements you want but not fully expressed, but extract the retrieval records you want from the massive database through computer artificial intelligence simulation analysis, judgment and integration.