3.2.1 IPC types

The IPC is the classification system of the WIPO and agreed to by its members. Every patent document published by a member has at least one IPC classification code.

Several classification systems are based on the IPC. This means that these classification systems have the same classification codes as the IPC with further subdivisions. These further subdivisions exist to make searching more efficient.

Well known classification systems based on the IPC are:

• Cooperative Patent Classification (CPC) used and maintained by the EPO and USPTO and also used by patent offices in other countries.
• File Index (FI) used by the JPO.
• Deutsche Klassifikation (DEKLA) used by the German patent office (DPMA).

As a hierarchical classification system the IPC type of classification systems require at least one classification code. A given code should be as detailed as possible. This means that a code that is on the lowest level possible should be assigned to a document. Multiple codes can be assigned to a single document, but the rules for assignment are not completely consistent between the different classification systems.

The IPC has eight sections:

1. Human Necessities
2. Performing Operations, Transporting
3. Chemistry, Metallurgy
4. Textiles, Paper
5. Fixed Constructions
6. Mechanical Engineering, Lighting, Heating, Weapons
7. Physics
8. Electricity

These eight sections are further divided into:

• Class
  • Subclass
    • Group
      • Main group or subgroups

As an example a classification code for a clothes peg with pivoting members: D06F55/02. This code consists of the parts:

The hierarchical structure with only the text of the specific headings of this example is:

To have a better impression of the hierarchical structure see the complete list of the classification of D06F in CPC scheme D06F.

In Characteristics is explained how the documents are classified. This knowledge of classification is necessary to be able to select the proper classification codes.

3.2.2 Indexing, deep indexing

Indexing classification systems will classify multiple aspects that are disclosed in a document. The multiple aspects that are used are for example features of a solution, application and problem. A single document has therefore multiple index classification codes.

The idea behind indexing classifications is that the contents of a document can be accurately defined by the multiple aspects and thereby also easily retrieved.

Although the CPC is a hierarchical classification system based on the IPC, some parts of the CPC also have an additional indexing classification system. This shows that a classification system does not have to be a pure hierarchical or a pure indexing system. As an example see the additional indexing codes (D06F2100 code series) in the D06F CPC in D06F.

The JPO uses both the FI a hierarchical system and the F-term system, a pure indexing system. The F-term system is a deep indexing classification system: multiple aspects are completely classified.

For clothes pegs the correct F-term is 4L027. However this F-term is empty and not used. A detailed scheme for clothes pegs exists in the FI, which is probably the reason why an F-term scheme for clothes pegs is not necessary.

For the technically neighbouring field for washing and drying machines a detailed F-term scheme exists. In F-term 3B168 this F-term scheme 3B168 can be seen. In this example multiple aspects can be classified:

• AA type of washer and dryer
• AB method of washer and dryer
• AC types of washer and dryer for businesses
• AD clothes
• AE description of drawings
• BA purpose
• CE rotor, churning wing
• FA detergent, solvent and medicine
• JM features of control
• WA instrument for wash and a wash auxiliary implement
• WB putting-in, taking-out and a carrying means of the washing

A document with a washing machine invention will get one or multiple classification codes from multiple aspects:

• characteristics of the type of washing machine and the process in the A aspects,
• purpose or problem solved in the B aspect,
• extra features in the C, F, J and W aspects if applicable.

A set of documents can be selected by combining the different relevant F-terms in a query. With this deep indexing it is thus necessary to classify all aspects, so that a good recall is possible. The aspects chosen in the classification scheme will have a big influence on the precision and recall. Also this scheme is built for examiners at the patent office and might therefore not fit properly on a research question.

An example of a commercial index classification is the Derwent manual codes classification system that is available in the Derwent World Patent Index (see website: clarivate). This is a proprietary classification system for searching in their patent database.

3.3.1 Introduction

To be able to choose the correct classification code and to be able to choose which classification system to use, the characteristics of the systems should be sufficiently clear. This part aims to make the most important characteristics clear.

The available databases have a limited number of classification systems available. The most used classifications systems are the IPC and the CPC. Therefore these two are compared here, with some small excursions to some other systems.

3.3.2 Scope of classified documents

All the members of the WIPO have to classify their patent publications with the IPC. This means that practically every patent publication has an IPC classification code.

Except that the IPC was introduced in the 1970’s. Documents published before this introduction could therefore not be published with an IPC code. In some databases, for example Espacenet, this is solved by assigning older documents an IPC code based on the available CPC codes on the document.

The EPO and USPTO only classify the documents from the most important countries with CPC codes. Documents in a language that is not an official language at the offices are nowadays not classified by these offices. For example documents in the Japanese language (both JP and WO) do not get CPC codes. Although not every document is classified with a CPC code, the number of documents with a CPC code is still huge. A quick check at the moment of writing reveals:

55 million of the total of 158 million documents in the database of the EPO have a CPC code assigned by the EPO or USPTO. A total of 70 million documents have a CPC code if the other countries that assign CPC codes are included. Compare this to the almost 37 million documents that have originally received an IPC code.

Numbers for other classification schemes: almost 16 million documents have an F-term, almost 18 million documents have an FI code. For an overview of the database at the DPMA see: data coverage Depatisnet.

Several countries (for example China and Korea) now also assign CPC codes to their published documents. It can therefore be expected that the coverage by the CPC will increase.

3.3.3 What is classified and how?

In both the IPC and CPC the invention is always classified. This means that a document has at least 1 classification code. It is also possible to classify additional information useful for searching. Thus multiple classification codes on a document are possible.

Generally the most detailed (lowest in the hierarchy) classification code is used. Another often used rule is that the first correct place in the classification scheme is used for the classification. But there are lots of exceptions. See the notes in the classification scheme and the CPC definitions for all the exceptions.

A very detailed guide on how to use the IPC can be found in the Guide to the International Patent Classification (2024). Most of what is written herein is also applicable to the CPC. This is not a document that can be read easily, so be warned.

The guide discloses the general rules. How in a specific technical field is classified with the CPC can be different than how generally is classified. The information how in a specific technical field is classified can be retrieved from the notes written in the classification scheme and from the definitions. See for example the definitions of D06F in CPC definition D06F.

The meaning of a group is often further explained in the definitions. If for example the difference between clothes pegs (D06F55/00) and clothes pegs with pivoted clamping members (D06F55/02) is not clear, the pictures in the definitions under these headings clarify this (see page 121 of the definitions in CPC definition D06F).

The CPC has more subgroups (see for example the parts in curly braces in CPC scheme D06F) and therefore the size of a selection can be smaller compared to using IPC codes. This will normally result in a higher precision.

Apart from more subgroups, the CPC has also extra additional codes (the 2000 range in subgroups). These additional codes are used in a variety of ways such as index or further subdivisions. A warning for this: additional codes are often not mandatory and thus may not be complete. Check what is written in the notes and definitions.

3.3.4 Who is classifying and when?

The IPC codes are assigned to a document, that will be published, by the examiner working at the specific patent office. Although effort is made for a consistent classification, it can be expected that there is some variety in the codes given by the examiners from the different offices. This can be seen by comparing for example the IPC codes with CPC codes on a document or comparing the IPC codes on the documents of the different countries in a patent family.

The CPC codes are assigned by examiners at the EPO or USPTO. These examiners work in a specialised technical field where a limited number of classifiers are active. This results in better consistency than the consistency with the IPC codes.

The IPC codes are only assigned when the document is published and are not updated.

The CPC codes assigned to documents are updated when an error is discovered or when the classification scheme is modified. The codes assigned to documents are therefore normally always up to date, whereby exceptions are mentioned in the classification scheme.

For the Dekla and the FI and F-term the situation is similar to the CPC.

3.3.5 Some guides

Experience will help to use the classifications correctly. So do not despair if the first time is not right.

Check by looking at the selected documents whether you have understood the classification correctly. Do not waste time looking through a wrong set!

If you can not find what you are looking for, you have made the wrong selection. Yes, prior state of the art exists for everything (unless you have made a world changing discovery).