DE102015112101A1 - Method for brainstorming - Google Patents

Abstract

For the development and improvement of products, in particular in the field of engineering design, a two-stage process for identifying relevant patent documents is proposed, in which natural persons are stimulated by the combination of the associated patent graphics (23) for brainstorming. For the automatic selection of the patent documents, a text classification is applied in a first stage and a picture classification on a previously compiled basic set (1) of patent documents in a second stage, whereby the associated text (3) and image classifiers (7) can be trained beforehand. In particular, the text classifier (3) can have a focus on content. The image classifier (7) may have a more formal character and, in particular, relate to the complexity of the patent graphics (23) to be determined.

Description

The invention is based on a method for brainstorming according to the preamble of independent claim 1.

Such methods can be used especially in the industrial environment to gain ideas for the development and improvement of products. In particular, the invention relates to the generation of innovations in the field of engineering design.

State of the art

It is for example from the document US5774833A known to examine texts and images from the patent literature in a computer with semantic methods.

The publication WO2008156507A1 describes a method for automatic patent evaluation by a computer program.

In order to increase competitiveness, internal processes for the development of innovative products continue to be established in many companies.

This is how the document describes US20120054281A1 a method of improving the performance of collaborative group innovation through team building in a virtual environment.

From the publication "A methodical way to innovative technologies" (author Spies, K., Verlag der Augustinus Buchhandlung, Aachen, 1996 ) it is known to present designers with selected images from patents of technical products, thereby enabling them to combine the technical principles presented there by restructuring and / or transferring them to other applications to new ideas. This approach has proven particularly advantageous in mining, mechanical engineering and in the construction industry, ie in areas of technical construction. It is disclosed in particular that the issued patent drawings not only from the respective area of responsibility, eg the IPC (International Patent Classification) Section / Class / Subclass / etc. but should instead cover as broad a thematic field as possible.

A disadvantage of this prior art is that the methods for innovation generation, especially in the field of small and medium-sized companies are often too expensive and continue to take against the background of ever-shorter product life cycles an undesirably large period of time.

task

The object of the invention is therefore to develop a cost-effective and time-saving method for innovation generation, in particular using a topic-based selection of patents from different technological fields of application.

The object is achieved by a method of the type mentioned by the features of the characterizing part of the independent claim 1.

• a.) automatische Analyse von Patentdokumenten, umfassend Patenttexte und Patentgrafiken, durch einen Computer;
• b.) automatische Auswahl relevanter Patentdokumente durch den Computer;
• c.) Darstellung der Patentgrafiken der ausgewählten Patentdokumente;
• d.) Betrachten und kombinieren von in den dargestellten Patentgrafiken enthaltenen Merkmalen zum Generieren neuer Ideen durch natürliche Personen;

wobei zur Auswahl der Patentdokumente in Verfahrensschritt b.) in einer ersten Stufe eine Textklassifikation mit einem Textklassifikator, aufweisend eine Textklassifikationsfunktion, und in einer zweiten Stufe eine Bildklassifikation mit einem Bildklassifikator, aufweisend eine Bildklassifikationsfunktion, durchlaufen werden.The brainstorming process includes the following steps:

• a.) automatic analysis of patent documents, including patent texts and patent graphics, by a computer;
• b.) automatic selection of relevant patent documents by the computer;
• c.) representation of the patent graphics of the selected patent documents;
• d.) viewing and combining features contained in the illustrated patent graphics to generate new ideas by natural persons;

wherein for the selection of the patent documents in step b.) in a first stage a text classification with a text classifier having a text classification function, and in a second stage an image classification with an image classifier having an image classification function are traversed.

Advantageously, in order to carry out the method, in particular steps a.) And b.), The computer executes a program stored in a corresponding memory of one or more processors likewise belonging to the computer, in particular one or more microprocessors.

Advantageous embodiments of the invention are specified in the subclaims.

The process is a method of quickly and inexpensively developing product ideas, particularly in the field of engineering design. Furthermore, the method is optimized in particular for improving corresponding, already existing products.

The method has the advantage that even small and medium-sized companies are supported in their brainstorming phase, because this is only a free or at least very inexpensive available basic amount of patent documents is needed and because the process continues with only a small financial and time effort and with comparatively little computational effort is connected.

The process thus has the advantage that such a brainstorming process can be carried out within a few days and thus meets the demand for short development periods. In particular, a.) The period of automatic analysis of the patent documents by the computer and b.) The period of automatic selection of the patent documents are extremely shortened, in particular by a suitable combination of the text classification with the image classification.

In an advantageous embodiment, the results of said two stages are linked to the Boolean operator "AND". This can be time-saving if at least parts of the method, for example for test purposes, are performed several times or if documents already read into the computer are repeatedly selected in different analyzes with the same classifier. For example, the result of the classification, e.g. the image classification, then stored as a classification function to the respective document quantity or the documents can be provided for example with a corresponding attribute.

If the method is expected to be carried out only once, it is particularly advantageous according to a further embodiment, when initially one of the two stages of patent documents to be examined is run through, and then only pass through the other of the two stages selected patent documents. In this way, not all patent documents need to go through both stages, saving time and computing capacity.

Since the second stage relates to the graphics evaluation and is likely to be more time-consuming than the first stage, it is particularly advantageous if the first stage, which concerns the text classification, is carried out in chronological order.

Thus, it is thus particularly advantageous if patent documents are first analyzed on the basis of their patent texts by means of the text classification and subjected to a selection, and then only the patent graphics of the selected patent documents are analyzed by the image classification and subjected to a further selection.

Thus, the second stage is only traversed by the patent graphics of the patent documents selected to be relevant in the first stage according to their patent texts, thereby saving considerable computing capacity and time. Finally, in this way, only a fraction of the patent documents to be analyzed in the first stage must pass through the particularly compute-intensive second stage. For example, a basic set may include 10,000 to 20,000 patent documents to be examined. In practice, the selection of the first stage may be, for example, about 5% -10% of the patent documents, ie, for example, 1000 documents, so that in the particularly compute-intensive second stage only these 1000 patent documents, i. by means of the image classification whose patent graphics must be examined, whereby in the second stage e.g. 90% -95% of the computational effort can be saved.

The classifiers, ie the text and image classifier, are formed as follows:
By means of a so-called "training" of a classifier, in particular of the text classifier, with manually selected training documents, the corresponding classification function, in particular the text classification function, is determined.

In a preferred embodiment, therefore, to train at least the text classifier for the current brainstorming project, thematically relevant patent documents, e.g. determined from a public database as training documents. For this purpose, a relevant search criteria can be used. In empirical experiments it has been found that a technical function, e.g. "Sawing", "Screwing", Welding "to corresponding devices, e.g. "Saw", "screw", "welding machine", achieved a much better hit rate.

From the set of hits, suitable patent documents can then be selected as training documents in the form of a manual final selection with little manual effort.

In a preferred embodiment, therefore, particularly relevant patent documents can be identified as relevant and used as training documents for training the text classifier.

Furthermore, non-relevant patent documents can also be identified as irrelevant and also used as training documents for training the text classifier.

In this case, for example, the feature known to the expert, so-called χ ² method ("Chi-Square method") Be used, which is described in more detail below:
By means of the χ ² method, the independence of two variables - and in this case the dependency of a feature on a category - in this case relevant / non-relevant - is generally determined. The χ ² method is therefore particularly well suited to finding the most meaningful features.

χ ² can be determined as follows:

A:

Anzahl der Dokumente aus einer Kategorie c, die ein bestimmtes Merkmal m enthalten;

B:

Anzahl der Dokumente, die nicht in der Kategorie c enthalten sind und, die das bestimmte Merkmal m enthalten;

C:

Anzahl der Dokumente aus Kategorie c, die das Merkmal m nicht enthalten;

D:

Anzahl der Dokumente, die nicht in der Kategorie c enthalten sind und die das Merkmal m nicht enthalten;

N:

Gesamtzahl der Dokumente in der dazugehörigen Trainingsmenge.

In this formula mean:

A:

Number of documents from category c containing a specific feature m;

B:

Number of documents that are not in category c and that contain the specific feature m;

C:

Number of documents from category c that do not contain the characteristic m;

D:

Number of documents that are not in category c and that do not contain m;

N:

Total number of documents in the corresponding training amount.

In general, the dependencies of all characteristics used can be calculated for all categories used and then averaged. The features can now be ranked for each category. The more dependent the characteristic of this category, the higher it stands in the ranking.

The use of the χ ² method has the further advantage that this already standardizes the interval [0, 1], which makes the results comparable.

In a particularly preferred embodiment, in the second stage, the image classification may use a formal feature, in particular the complexity of the patent drawings, as a selection criterion. This is particularly advantageous for several reasons:

On the one hand, a corresponding image classification function after a single creation can be used for all such methods, and thus can be regarded as generally prescribed for such methods independently of the respective topic of the idea generation project.

Training the image classifier in this case only needs to take place once. For this, a set of training documents with manually selected, complex and less complex graphics is assembled once, whereby the individual graphics for training the image classifier are marked as complex or less complex. Then the image classifier can be calculated once, eg with the said χ ² method.

On the other hand, this criterion of the complexity of the graphics for the natural persons, so for example developers, designers, potential inventors, amateurs interested, etc., also particularly advantageous in the interpretation, since in particular the manual recognition by the selection of less complex graphics significantly improved and thus significantly reduces the manual intellectual effort involved in combining the elements contained therein.

The complexity of the graphics can be determined, for example, by the ratio of their lines to the total area of the graphic. It is advantageously to be considered that patent drawings are basically line drawings, and that the strength of the lines is comparable among a large number of suitable patent documents. For example, the length and / or area of the lines may be proportionate to the size, i. Diagonal or plane, the overall graphics are set. In particular, the number of corresponding pixels can be used.

Thus, the proportion of lines of current analysis tools, e.g. Image analysis software, easily recognized and evaluated with low computational effort.

Furthermore, patent graphics in which the proportion of lines on the overall graphic is below a certain predetermined value can be considered to be less complex. Patent graphics in which the proportion of lines in the overall graphic is greater than the predetermined value can be considered complex.

The said predetermined value can, in an advantageous embodiment, also be determined on the basis of the entire image material to be examined, e.g. be determined as the mean. Then automatically the least complex graphics in relation to the whole are selected.

The value can also be fixed by the method independently of the image material to be examined. For example, this value, for example based on the ratio of the areas, may be relatively small and may be 0.01% or 0.05% or 0.1%. but it can also be at 0.25% or 0.5%, 1% or 2.5% or else at 5% or 7.5% or 10% or more, eg at 15% or 20%. or even above, eg at 25%, 30%, 35% or even 40%.

This value can also be used to set absolutely absolute, how complex the image material may be in order to be part of the selection.

Thus, the image classification can use the complexity of the patent graphics as a selection criterion by selecting those patent documents in the patent graphics of which the ratio of lines to the total area of the graphic is less than those patent documents which are not selected.

In this way, the less complex patent graphics can be automatically selected according to a predetermined criterion with little computational effort.

In a further advantageous embodiment, the image classifier, analogous to the text classifier, can be generated by a selection of content-relevant and non-relevant and appropriately marked training documents. This may be useful if the images of relevant documents have significant similarities, which may be, e.g. refer in particular to the topic focus of the brainstorming project according to the IPC and / or CPC classification or comparable further classifications.

It is thus possible that the image classifier in a particularly advantageous embodiment also has content-related components, that is, for example, for certain geometric shapes, e.g. special so-called "mating faces", locking devices, circuit arrangements and associated symbols, etc. searches and selects corresponding patent documents. Thus, in an advantageous embodiment, the image analysis may also include, for example, pattern recognition and / or pattern analysis methods.

There are several ways to train and apply the text classifier / classifiers:
A first approach is to use patent documents of various such sections, in particular IPC or CPC sections, separately in the training phase as training documents so as to obtain a separate text classifier for each section. In the subsequent selection phase then patent documents of the various sections can be selected with a respective section-specific text classifier.

Although this means a much higher configuration and computational effort, but it seems at first quite understandable that, for example, the so-called "Precision", which specifies the ratio of number of relevant documents / number of documents found, so in simple words "the accuracy" of each classifier indicates that, at least because of the uniform meanings of words within a section, this would have to significantly improve.

Surprisingly, this thesis was not confirmed at least by the test series carried out. At least in the investigations carried out no appreciable influence could be found by the use of IPC-section-specific text classifiers.

It has therefore proven to be particularly advantageous for the reduction of the manual effort and / or the overall considerable computational effort to pursue an alternative approach, which consists in, in the first stage, i. in the text classification, to select patent documents from different IPC / CPC or possibly also from other patent classification sections by means of the same text classification function. In other words, a single classifier for patent documents from different sections, e.g. ICP / PCP sections, without significantly degrading the result. As a result, both the manual effort and the computational cost of the computer can be significantly reduced both when configuring the system and during the actual regular choice of patent documents.

For the selection phase, when assembling the basic set of patent documents from a database, e.g. from a public database, for example with the term "locking" not only from the IPC section H "electrical engineering" and section F "mechanical engineering", but also in section B "working methods" and section A "Daily life requirement" patent documents are compiled. It is therefore also possible to select documents relevant to the text and / or the image classifier from the basic set composed in this way. The text classifier can of course have a stronger content component and the image classifier can preferably have formal criteria, such as e.g. to evaluate the said complexity.

The representation of the patent graphics for viewing and combining by natural persons, eg developers, designers, Experts, interested amateurs, etc., are by no means limited to conventional exhibitions in the form of galleries in which printouts of the graphics can be posted conventionally. In a further preferred embodiment, these graphics can be distributed, for example via a network, to computers of the subscribers, ie the said natural persons, and displayed there in the form of a program, eg as a slide show. The order of the individual graphics, eg depending on a random number generator, can vary for different subscribers. Also, the same graphics may be presented to the same participants multiple times but in a different order to trigger different associations and create combinations. Furthermore, embodiments are conceivable in which the participants can select the computer monitor continuous graphics by mouse click or parts of it can combine and save.

embodiment

An embodiment of the invention is illustrated in the drawings and will be explained in more detail below. Show it:

1 a simplified procedure of a two-stage selection phase;

2 a more detailed process of the two-stage selection phase;

3 a process diagram of an associated training phase;

4 a flow of an entire brainstorming process;

5a a classification of patent documents of different sections by means of several section-specific classifiers;

5b a classification of patent documents of different sections by means of a common classifier.

The figures contain partially simplified, schematic representations.

In part, identical reference numerals are used for the same but possibly not identical elements. Different views of the same elements could be scaled differently.

The 1 shows a basic, roughly simplified process of a two-stage patent classification in a so-called "selection phase". The term "selection phase" serves to distinguish it from a so-called "training phase" otherwise similar in the process and means that the relevant patent graphics determined in the selection phase represent a regular result that is later used to find ideas by natural persons, eg developers, designers , Professionals, interested lay people, etc., can be spent.

In this two-stage patent classification, first, a first stage, comprising a text classifier 3 , of patent documents of a basic quantity 1 for the analysis of the associated patent texts 2 go through and then becomes a second stage, having an image classifier 7 only of the patent documents selected in the first stage for the analysis of the associated patent graphics 6 run through.

The basic quantity 1 theoretically could consist of hundreds of thousands or even millions of patent documents. However, as a result, only about 100 relevant patent graphics were used for the brainstorming method 23 can be required, a much lower basic amount 1 be sufficient, for example, 5000 to 10,000 documents.

To obtain the basic quantity 1 From a public database, for example, a thematic preselection, eg by a keyword search and / or a rough restriction of the IPC / CPC sections or associated patent classes, may have already taken place. This is particularly advantageous because the timely avoidance of completely irrelevant subject complexes saves a great deal of computing power, which makes the method much more efficient.

In the first stage, the patent texts 2 the patent documents of the basic quantity 1 with a text classifier 3 , having a text classification function γ selected. This will be out of the basic set 1 a first subset 4 textually relevant patent documents selected. At the same time, this also automatically creates a residual quantity 5 textually irrelevant patent documents generated, which is no longer considered in the further process.

The second stage is followed by the patent graphics 6 which are among the patent documents of the first subset 4 belong, compiled. The patent documents of the first subset 4 are now by a second classification, namely an image classification with an image classifier 7 comprising an image classification function ε, again divided into two further so-called "categories", namely a second subset 8th and in another remaining quantity 9 , where the second subset 8th now includes those patent documents which, both according to their textual content and according to their patent graphics for manual Analysis, combination and / or brainstorming are suitable. These patent documents are in this context as relevant documents and their graphics accordingly as relevant patent graphics or as relevant patent images 23 designated. The remaining amount 9 , formed from the patent documents of the first subset that are not relevant according to their graphics 4 , is no longer considered in the present process.

The 2 shows a more detailed process diagram of the selection phase in the two-stage process.

In the first step 11 become the patent texts 2 the basic amount 1 , for example, in the usual XML format, read. In one possible embodiment, this is done in the form of full texts, comprising the so-called "abstract", ie the abstract, furthermore the so-called "prior art", the so-called "patent description", the so-called "embodiment" and the so-called "claims". Preferably, however, the text section which describes the prior art can be dispensed with because, strictly speaking, it does not describe the invention itself. Furthermore, the reading-in of the claims can advantageously also be dispensed with because they have increasingly juristically embossed formulations and terms which, according to experience, are less well suited for brainstorming. In this case, the read-in texts thus comprise only the abstract, the general patent description and the exemplary embodiment. For this purpose, it is particularly advantageous if the text format, this distinction allows, that identifies the different text sections accordingly.

At the second step 12 An extraction of features takes place. This can be done by finding complete terms or by partial terms, ie so-called "n-grams". Thus, for example, the following trigrams (N = 3) are formed from the sentence "They walk to their homes": "they, lukewarm, open, close, to, their, oh, ah, hau, out, use". Other feature expressions are possible in this context, for example, consisting of several terms with defined word spacing up to the grammatically defined sentence structures with a subject / predicate / object analysis. In connection with this, or separately from it, it is also possible to merge synonyms in a meaningful way and, depending on the task, may prove useful. Furthermore, a previous use of a so-called "stop-word filter", which eliminates meaningful terms such as articles, conjunctions, prepositions, etc., as well as the use of a thesaurus for correcting possible spelling errors, in particular by scanning documents and a subsequent optical character recognition, such as the so-called "OCR" (optical character recognition), can arise, possible and useful. It can find ² method using also said χ.

In the feature weighting in the third step 13 The frequency of the respective characteristics can be used absolutely or normalized and in particular also with their total frequency in the basic quantity 1 be matched.

According to a so-called "retrieval model", these weighted features can be analyzed and in the fourth step 14 by using the text classification function γ.

The usual retrieval models used in practice are usually the so-called "Boolean model", the so-called "vector space model" and the so-called "probabilistic model":
These are characterized as follows:
The Boolean model, also known as "keyword search", is based on the approach of searching text documents for the presence or absence of keywords. The application searches by means of single words. The result of this search is a set of documents containing these keywords. The search terms can be combined by the logical operators "AND", "OR" and "NOT". A ranking of the result set is not possible.

The vector space model is based on mapping both the query and the documents of the result set as vectors in a high-dimensional space. The vectors of the query and each result are compared. The more similar these vectors are, the more highly the relevancy of the particular document of the result set for the response to the query is estimated. This results in a ranking of the documents in the result set. To use the vector space model for the classification of documents, for example, the so-called "Support Vector Machine" can be used.

The probabilistic model is based on probability values. The biggest challenge in so-called "information retrieval" in texts is the vagueness of the language. Thus, there is no absolute assurance that a document is relevant to a request. Therefore, probabilities for the relevance of documents are calculated. Here, the relevance is given as a similarity value. The similarity depends on the frequency of search terms in the document. The higher the calculated probability, the more relevant the document is estimated for the query. As a probability-based classification method can For example, the so-called "Naive Bayes classification method" find use.

According to their weighting, the patent documents can by the fourth step 14 applying the text classification function γ to their patent texts 2 in the fifth step 15 be assigned to a category. The category is in particular the one described above and in 1 illustrated first subset 4 ,

To the way in the sixth step 16 relevant patent documents found in the seventh step 17 the associated patent graphics 6 compiled.

In the eighth step 18 become these patent graphics 6 , for example, as PNG files or in any other graphics format read. The graphics of the textually irrelevant patent documents of the remaining quantity 5 will not be read. This avoids the import of content-irrelevant graphics and thus leads to a lean and resource-saving process.

In the ninth step 19 a feature extraction is performed from the patent graphics with a so-called "image mining".

Analogous to the text analysis, digital images, eg the patent graphics, can be used in image mining 6 the patent documents of the first subset 4 to be targeted according to given criteria. This is based on a content-based image search, namely the so-called "Content Based Image Retrieval". The content of an image is analyzed by a software algorithm. For example, colors, outlines, and textures are used. With the aid of various methods, features can be extracted from the image and displayed as feature vectors. As soon as the image information is available as feature vectors, the methods for feature selection and in the tenth step for feature weighting can now be used analogously to the text analysis described above 20 be applied.

So by the eleventh step 21 successively applying the image classification function ε the patent graphics in the twelfth step 22 different categories 8th . 9 be assigned to. In particular, a binary classification takes place, ie the graphics are split according to a yes / no decision between two different categories, namely between the second subset 8th and the associated further residual quantity 9 , The second subset 8th then includes the relevant patent documents, including the relevant patent images 23 , which are suitable for manual combination / analysis / brainstorming by natural persons.

In order to be able to perform such selection methods, however, the text classification function γ and the image classification function ε must first be calculated.

The 3 shows the procedure for training the text 3 and the image classifier 7 , ie for the calculation of the respective classification function γ, ε.

These are used to train the classifier, in this case the text classifier 3 , first manually selected substantively suitable patent documents, which are referred to below as training documents. These training documents should not be part of the basic quantity 1 so as not to distort the result by so-called "over-training". For this purpose, these training documents, for example, first searched with keywords in a public patent database, for example, and then manually selected. As a guideline, for example, at least 100, so for example about 250-500 documents can be used in this way as relevant training documents. Furthermore, a lot of non-relevant training documents are selected and selected in about the same number, ie also at least 100, preferably 250-500 documents. These manually selected relevant and non-relevant training documents are used below as training documents.

In the first step 11' the training of the text classifier 3 The texts belonging to these training documents are read into the computer and marked as relevant or irrelevant.

At the second step 12' finds an extraction of features from these training documents analogous to the second step described above 12 the selection phase. In this case, the dem ² method known to those skilled in the art and already described in detail above can be used as a simplified approach for the probabilistic model on the basis of frequency values instead of probabilities. There is also a considerable saving in computing power for the following reason:
Because many features would greatly slow down the process in both the training phase and the selection phase, it is advantageous to use only the most meaningful features, those features that most clearly distinguish the relevant from the irrelevant documents. By means of the χ ² method, the independence of two variables - and in this case the dependency of a feature on a category - in this case relevant / non-relevant - is generally determined. The χ ² method is therefore particularly well suited to finding these most meaningful features.

The χ ² method can be used for feature extraction 12 . 12' . 19 . 19' be used both in the training and in the selection phase for both text and image analysis. Furthermore, the χ ² method can be used for the calculation 30 . 31 the classification functions ε, γ are used.

In the third step 13' the training phase finds the feature weighting 13' analogous to the feature weighting 13 the selection phase. In particular, the frequency of occurrence of particular identifiers can be evaluated and can furthermore be set in relation to the total frequency of these identifiers.

In the fourth step 14' the text classification function γ is calculated. In this case, for example, with the χ ² method described above, the relevance of individual weighted features can be used in order to select the most relevant features in relation to the searched content.

The training of the image classifier 7 is essentially analogous to the aforementioned training of the text classifier 3 , A suitable number> 100, eg between 250 and 500 relevant graphic documents are selected manually as training documents according to their properties. These documents should not be preferred to the patent documents of the basic set 1 and especially not to the first subset 4 belong to not distort the result by the said overtraining.

These training documents are in the first step 18' the training of the image classifier 7 marked as relevant or not relevant and read into the computer.

At the second step 19' the features are extracted by means of the above-mentioned "image mining" and in the third step 20' weighted eg with the χ ² method, so in the fourth step 21' to be able to calculate the classification function ε.

As an alternative to training the image classifier 7 This can also be fixed and can be part of the procedure. In particular, this makes sense if the image classifier 7 refers exclusively to formal features, such as the complexity of the graphics. Either he can then be determined once by training with training documents selected according to these formal criteria. Alternatively, a fixed value can be defined as a criterion, for example the ratio in the graph of recognized lines, eg their total length or their area, eg number of pixels, to the overall dimension of the graphic, eg its diagonal or area, eg number of pixels. In addition, the complexity of the searched relevant patent graphics is then manually adjustable.

The 4 exemplifies a possible overall flow of a method for brainstorming.

In point I .) becomes the basic quantity 1 of at least 500 patent documents with associated patent texts 2 a specific language.

In point II .) takes place independently of a manual selection of training documents depending on the topic of the technical problem. The training documents should not be in the basic quantity 1 be included.

In point III .), the training of a particular binary text classifier takes place 3 with training documents by means of an automatic feature extraction from the training documents.

In point IV .) becomes the text classifier thus determined 3 on the patent documents 2 the basic amount 1 applied. This will become the first subset 4 generated.

In point V .) become patent graphics 6 to the patent documents of the first subset 4 compiled.

In point VI .) manually become suitable patent graphics as training documents for the image classifier 7 The associated patent documents preferably do not belong to the basic quantity 1 and especially not to the first subset 4 belong. The patent graphics may be relevant and irrelevant and labeled accordingly for training the image classifier 7 be used.

In point VII .) becomes an image classifier 7 through the automatic feature extraction 19' trained from these training documents. Alternatively, a given image classifier can also be used 7 In particular, a formal and therefore cross-thematic criterion, for example the complexity of the images, is used as a selection criterion.

in point VIII .) becomes the image classifier 7 on the patent graphics 6 the first subset 4 applied to a second subset 8th to create.

In point IX .), the patent graphics of the patent documents become the second subset 8th output.

In point X .) can optionally provide technical solutions in the patent graphics of the second subset 8th are combined with each other. This point is not necessary for the implementation of the procedure

In point XI .) can the patent graphics or combination of solution blocks from the optional point X .), For example, as images are output, for example, by visibly unhooking the prints in a gallery or by graphical representation using a computer program, for example via a network or the like.

In point XII .), the combinations are evaluated manually by the natural persons, ie participants of the brainstorming process.

The 5 represents the difference between the selection by means of in 5a shown section-specific text classification functions γ ₁ , γ ₂ , γ ₃ and the in 5b represented non-section-specific text classification function γ.

In the 5a the patent texts of individual IPC sections S1, S2, S3 are each provided with an associated text classifier 3 ' . 3'' . 3''' , having a corresponding text classification function γ ₁ , γ ₂ , γ ₃ , selected.

From the respectively determined relevant patent documents are then each subset quantities 4' . 4'' . 4''' generated. These subset quantities 4' . 4'' . 4''' can then go back to the first subset 4 to be united. The non-relevant patent documents are according to the associated remainders 5 ' . 5'' . 5''' fed, which are no longer considered for the process.

In the 5b is shown as the patent documents of the three different IPC sections S1, S2, S3 on a single common text classifier 3 for generating the first subset 4 to be selected.

In practice, this selection works through a single common text classifier 3 According to the present test results faster and without significant qualitative losses compared to the much more cumbersome section selection by several text classifiers 3 ' . 3'' . 3''' ,

LIST OF REFERENCE NUMBERS

1

 Basic set of patent documents

2

 Patent texts of the patent documents of the basic quantity

3, 3', 3'', 3'''

 Text classifier, section-specific text classifiers

4

 first subset

4', 4'', 4'''

Subset sets of the first subset

5, 5', 5'', 5'''

 Remaining amount (n)

6

 Patent graphics of the patent documents to be analyzed

7

 Bildklassifikator

8th

 second subset

9

 further residual quantity

11

 Reading the patent texts of the basic quantity into the computer

11'

 Reading the training documents into the computer

12, 12'

 Feature extraction (text)

13, 13'

 Feature weighting (Text)

14

 Apply the text classification function

14'

 Calculation of the text classification function

15

 Assignment of relevant patent documents to the first subset

16

 relevant patent texts

17

 Compiling relevant patent graphics

18

 Reading in for patent graphics

18'

 Reading the patent graphics of the training documents

19, 19'

 Feature extraction (graphic)

20, 21

 Feature weighting (graphic)

21

 Apply the image classification function

21'

 Calculation of the image classification function

22

 Assignment of relevant patent graphics to the second subset

23

 relevant patent graphics

γ, γ ₁ , γ ₂ , γ ₃

Text classification function, Section-specific text classification functions

ε

 Image classification function

S1, S2, S3

 Patent texts of individual patent certification sections (e.g., IPC sections, CPC sections)

I) -. XII).

Process steps of a possible overall sequence of a method for brainstorming

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• US 5774833A [0003]
• WO 2008156507 A1 [0004]
• US 20120054281 A1 [0006]

Cited non-patent literature

• Spies, K., Verlag der Augustinus Buchhandlung, Aachen, 1996 [0007]

Claims (10)

Method for generating ideas, in particular for developing and improving products in the field of technical construction, comprising the following steps: a.) Automatic analysis of patent documents, comprising patent texts ( 2 ) and patent graphics ( 6 ), by a computer; b.) automatic selection of relevant patent documents by the computer; c.) Presentation of the patent graphics ( 23 ) of the selected patent documents; d.) viewing and combining features contained in the illustrated patent graphics to generate new ideas by natural persons; characterized in that for the selection of the patent documents in method step b.) in a first stage a text classification with a text classifier ( 3 ), comprising a text classification function (γ), and in a second stage an image classification with an image classifier ( 7 ) having an image classification function (ε). A method according to claim 1, characterized in that the results of the two stages are combined with the Boolean operator "AND". A method according to claim 1, characterized in that initially one of the two stages is run through, and that then only the patent documents selected thereby pass through the respective other of the two stages. Method according to claim 3, characterized in that first of all the first step of 1 patent documents, with part of the patent documents being considered as a first subset ( 4 ), and that then the second stage is only covered by the patent documents of the first subset ( 4 ). Method according to one of the preceding claims, characterized in that the text classification function (γ) is obtained by training the text classifier ( 3 ) is determined with manually selected training documents. Method according to claim 5, characterized in that said training documents are previously determined manually on the basis of a search criterium relevant to the respective brainstorming, in particular on the basis of a technical function, and identified in each case as relevant or not relevant. Method according to one of the preceding claims, characterized in that the image classification the complexity of the patent graphics ( 6 ) used as a selection criterion. A method according to claim 7, characterized in that the image classification function (ε) is predetermined. Method according to one of the preceding claims, characterized in that the image classification function (ε) by training the image classifier ( 7 ) is determined with manually selected training documents. Method according to one of the preceding claims, characterized in that in the text classification patent documents from different sections (S1, S2, S3) are selected with the same text classification function (γ).

Images