What is it?


Sophia Query is an AI chat box assistant, trained to take your Natural Language questions and turn them into a Boolean Patent Search Query.  It allows you to chat à la ChatGPT to create a query. Its sole purpose is to help you build a Boolean query to find the most relevant documents. The aim is not to be exhaustive but focused. 



How to use it? 


From the Easy Search page click on the stars at the right corner of the search ribbon.



A new pop up window will open. Just start chatting with Sophia Query to simply begin building your query. 



Type what you want to search for, such as “Find patents about drones and batteries” and the chat box will automatically add synonyms and IPC classes to build a query.  

You can adjust the query by asking the bot to remove or add keywords or classes.  


You can use any language you want to write your question but the assistant will only build an English answer. 

You don't have to worry about misspellings errors, the assistant can handle it.


You can refine your search by asking about authorities (e.g., “find patents from EPO and the US”), inventors, and assignees (e.g., “find patents from Ian Fleming” or “find patents from Tesla”), publication dates and filing dates (e.g., “Find patents published after 2015 and filed before 2020”).  It also recognizes patents status such as alive or granted. 
You can also precise if it is an FTO and the bot will focus on claims and alive patents!


Some question's examples:

  • Find patents about drones and batteries published before January 1st, 2023, and published by WIPO.
  • Make IPC classes more precise.
  • Remove the keyword enzyme and IPC C12N.
  • Add the top 5 companies in this domain.
  • Find patents for IoT devices published in China before 2015 and filed after 2000.


The patent's elements Sophia Query recognizes are:

  • Text fields (generic text and claims)
  • Authorities 
  • Assignees and inventors
  • IPC and CPC classes
  • Filing/application and publication dates
  • Status : alive / granted

Asking about any other fields will not work. More fields will be added soon.


To display patent lists, you can launch your query by clicking on the Search button.


You can always see and modify the Boolean query generated in the Command Line Editor clicking on Modify query button, but note that this change cannot be detected by the chat bot and is therefore ignored. 




For a new conversation, you should click the Reset chat button to ensure that the AI bot is not influenced by previous requests. 


What will not work: Unknown fields, vague queries, questions impossible to answer in a Boolean query, or not related to patents.

Examples:

  • Find patents from best companies. (too vague)
  • Are WIPO patents any good? (not related to querying for patents)
  • How many patents are from China? (numerical issues are not handled)
  • I want a recipe for ratatouille! (unrelated to patents)


Challenges:

  • Sometimes, IPC classes are erroneous. Safeguards and double checks will be implemented soon.
  • Keywords and synonyms are too long.
  • Multi-lingual capability is not perfect.
  • Asking several times the same question provides slightly different answers. This is a phenomenon common to AI systems.


Note: You cannot ask questions unrelated to patent querying, though the chat bot will still be quite flexible. For instance, asking for a lasagna recipe will not be accepted, but rather add the top 10 pharma assignees to your query.



A typical conversation :


Question 1: 

Find patents for CRISPR from Caribou Biosciences 

Answer:

(CRISPR OR (CRISPR-Cas9) OR (gene 3W editing) OR (genome 3W editing) OR (clustered 3W regularly 3W interspaced 3W short 3W palindromic 3W repeats))/TI/AB/CLMS/SA AND ((Caribou 1D Biosciences))/PA/OPA/NPAN AND (C12N OR C12Q OR C12P)/IPC 

Interpretation:

  • CRISPR synonyms: "CRISPR-Cas9", "gene editing", "genome editing", "clustered regularly interspaced short palindromic repeats"
  • Caribou Biosciences is a company known for its work in CRISPR technology.
  • Relevant IPC classes for CRISPR: C12N (Microorganisms or enzymes; compositions thereof), C12Q (Measuring or testing processes involving enzymes or microorganisms), C12P (Fermentation or enzyme-using processes to synthesise a desired chemical compound or composition or to separate optical isomers from a racemic mixture).


Question 2:

Applied to autoimmune diseases

Answer:

(CRISPR OR (CRISPR-Cas9) OR (gene 3W editing) OR (genome 3W editing) OR (clustered 3W regularly 3W interspaced 3W short 3W palindromic 3W repeats))/TI/AB/CLMS/SA AND ((autoimmune 3W disease) OR autoimmunity OR (immune 3W disorder))/TI/AB/CLMS/SA AND ((Caribou 1D Biosciences))/PA/OPA/NPAN AND (C12N OR C12Q OR C12P OR A61K)/IPC

Interpretation:

  • Autoimmune disease synonyms: "autoimmunity", "immune disorder"
  • Relevant IPC classes for autoimmune diseases: A61K (Preparations for medical, dental, or toilet purposes)


Question 3 :

Change CRISPR-Cas9 as CAS9

Answer :

(CRISPR OR CAS9 OR (gene 3W editing) OR (genome 3W editing) OR (clustered 3W regularly 3W interspaced 3W short 3W palindromic 3W repeats))/TI/AB/CLMS/SA AND ((autoimmune 3W disease) OR autoimmunity OR (immune 3W disorder))/TI/AB/CLMS/SA AND ((Caribou 1D Biosciences))/PA/OPA/NPAN AND (C12N OR C12Q OR C12P OR A61K)/IPC

Interpretation :

CRISPR synonyms updated: "CAS9", "gene editing", "genome editing", "clustered regularly interspaced short palindromic repeats"


Question 4:

Add the list of top 5 autoimmune diseases 

Answer:

(CRISPR OR CAS9 OR (gene 3W editing) OR (genome 3W editing) OR (clustered 3W regularly 3W interspaced 3W short 3W palindromic 3W repeats))/TI/AB/CLMS/SA AND ((autoimmune 3W disease) OR autoimmunity OR (immune 3W disorder) OR (rheumatoid 3W arthritis) OR (type 3W 1 3W diabetes) OR (multiple 3W sclerosis) OR lupus OR (inflammatory 3W bowel 3W disease))/TI/AB/CLMS/SA AND ((Caribou 1D Biosciences))/PA/OPA/NPAN AND (C12N OR C12Q OR C12P OR A61K)/IPC

Interpretation:

Autoimmune diseases included: "rheumatoid arthritis", "type 1 diabetes", "multiple sclerosis", "lupus", "inflammatory bowel disease" 


Question 5:

Filed after November 2010

Answer:

(CRISPR OR CAS9 OR (gene 3W editing) OR (genome 3W editing) OR (clustered 3W regularly 3W interspaced 3W short 3W palindromic 3W repeats))/TI/AB/CLMS/SA AND ((autoimmune 3W disease) OR autoimmunity OR (immune 3W disorder) OR (rheumatoid 3W arthritis) OR (type 3W 1 3W diabetes) OR (multiple 3W sclerosis) OR lupus OR (inflammatory 3W bowel 3W disease))/TI/AB/CLMS/SA AND ((Caribou 1D Biosciences))/PA/OPA/NPAN AND (C12N OR C12Q OR C12P OR A61K)/IPC AND (EAPD > 2010-11-01)

Interpretation:

EAPD > 2010-11-01 added to the query


Disclaimer: 

The feature utilizes OpenAI/ChatGPT on public data via Questel servers. No personal information is shared with third parties.
ChatGPT may generate inaccurate information regarding people, places, or facts.