[tweetmeme source=”Intellogist” only_single=false]
Patent classifications are the original patent searching tools and remain among the most useful ways to categorize and index related prior art. Classifications, by definition, are groups of patent documents of similar subject matter. Several major patent classification systems exist, including US, IPC, ECLA, Japanese F-Index and F-Term, and DEKLA. In previous posts, we have discussed how patent searching by classification works and why it’s worthwhile. To check out those posts, see parts one and two of the “Classified Information” series.
Patent classifications can be vitally important in conducting a patent analysis project, and today we’ll show you how to utilize classification systems to pare down your pool of patent documents under review in addition to potentially finding new documents of interest.
Last time, in Learn Practical Patent Analysis: A Case Study (Installment 2), we discussed how manual de-duplication can reduce the redundant patent documents returned to us in our first full text search query. Barring that, we discussed how patent families could be used as a de-duplication tool to limit certain closely related patent applications from overwhelming the project: removing family duplicates in the data set is desirable because it reduces the amount of time needed to perform both manual and statistical analysis processes, and it focuses the analysis on the inventive concepts rather than the patent documents. To keep this case study concise and to the point I’ve chosen an automatic family de-duplication process to represent each patent family as one object. The analysis that follows is based on this method, but different patent analysis studies require different methods, a point not lost on our commenters last week.
It is possible to run a statistical analysis on the 5000 or so references we have previously gathered in our solar panel search in many major patent search systems. One alternative option is to use the data file we exported and manipulate the data in Excel or another spreadsheet program. The best alternative option is to create custom “patent classifications” by hand by reviewing all references deemed to be relevant. This process is time consuming, but produces the best and most tailored results considering the client and subject matter at hand. If we were doing a more routine or shortened patent search, we might wish to turn take a slice of the selected patent documents and run an analysis on said section, but since this case study is focusing on a patent analysis project we will run a statistical analysis on the first 5000 documents. The bigger the sample size is, the less likely it is for outliers to affect our analysis.
Choosing the correct patent classification system is another issue which will arise during the analysis process. It is our recommendation that at least both main classification systems (US and IPC) be analyzed separately and included in the final report documentation, since each has their own strengths, weaknesses and idiosyncrasies. The relative strength between the US and IPC classification schemes may differ depending on the technology area. US classification tends to be narrower and more specific, while IPC classification tends to have a better hierarchical structure. ECLA, on the other hand, may be worth examining because of the high level of upkeep, ensuring that the classifications keep pace with the relevant technology. Also, keep in mind cases such as “patent families without a US patent family member” will not have any associated US classification data. For the purpose of this post, I’ll focus on analyzing the IPC classifications.
Looking at the statistical analysis of our first 5000 patent families, we see that the following are the top three IPC Sub-Classifications:
- F24J 2/00: Mechanical Engineering/Heating/Production or use of heat not otherwise provided for/Use of solar heat, e.g. solar heat collectors – 1682 hits
- H01L 31/00: Electricity/Basic electric elements/Semiconductor devices/Semiconductor devices sensitive to infra-red radiation, light, electromagnetic radiation of shorter wavelength, or corpuscular radiation and specially adapted either for the conversion of the energy of such radiation into electrical energy or for the control of electrical energy by such radiation – 835 hits
- B60H 1/00: Performing Operations/Vehicles in general/Arrangements or adaptations of heating, cooling, ventilating, or other air-treating devices specially for passenger or goods spaces of vehicles/Heating, cooling or ventilating devices – 355 hits
We can see right away that F24J 2/00 is exactly what we’re looking for—solar panels are one type of solar heat collector. The second IPC classification deals with the semiconductor technology that may be used to convert solar energy to electrical energy, but for the purposes of this case study we’re going to stick solely with the mechanical and structural facets of solar panels. The third IPC classification is somewhat of a false positive, focusing on vehicle HVAC systems (that may use solar panels for power supply or may just be extraneous hits). There may be good references in these categories, but it’s likely these good references will also be tagged with more relevant classes as well. Any time you discard data, there is a risk that useful prior art will be lost, and the bigger and more important the project is, the bigger and more inclusive you want to make your “net” when filtering prior art such as we are now.
That being said, for the sake of simplicity we are going to choose F24J 2/00 and all of its subclasses (which can be found at the WIPO Reformed IPC Internet Publication) to filter our 5000+ patent documents. In the search system we’re using, this can be done by appending a wildcard to the classification search after the “2/” in order to capture 2/00, 2/02, 2/04, etc. and then combining this string with our previous string that yielded 5000+ results. The end result is a narrowed down field of 1786 results to further analyze. At this point, if the subject matter was more narrow than “solar panel,” deeper subclasses within the IPC directory would be worth looking into and analyzing.
It’s important to note that a list of classifications by frequency that includes the H01L 31/00 and B60H 1/00 classifications among many others may be useful to your client. Graphical representations such as pie charts or heat maps may also be useful—we’ll touch on graphical presentation later on in the series. Cataloging and making note of this data is important to give a broader picture of what classifications this subject matter is being patented in and is useful information in its own right. If we started with a narrower initial search this would be even more apparent, but for our current broad search it is useful to see that, for example, a top 10 IPC classification for this art is E04D 13/00: Fixed Constructions/Buildings/Roof coverings/Special arrangements or devices in connection with roof coverings. Documenting this, via table, graph, or written up in a report shows our client that this classification area may be important for them either to investigate their competition or to show a technology area where their invention may prove useful and marketable. In this case perhaps their solar panel technology can be adapted into a roof covering form. Again, this will all rely on the scope and nature of the patent analysis project, but keeping such things in mind will help you consider how to best serve your client—or yourself, as the case may be!
Next time we’ll discuss how to determine the most relevant companies in this technology field within our newly revised data set.
What’s your favorite classification system to use? How granular do you think we need to go? If you have any questions about this case study thus far, feel free to leave us a message in the comments below!
Read installment 4.