Google Patents Scraper

This Actor extracts data from Google Patents.

It is designed for scraping patent/publication numbers directly or finding patents by keyword and returning structured metadata in an Apify dataset.

---

What it extracts

For each patent or publication number, the Actor can extract:

• Patent number (when determinable)
• Publication number
• Application number
• Title
• Inventor
• Assignee
• Legal status
• Priority date
• Publication date
• Expiration date (including adjusted expiration where available)
• Abstract
• Claims
• Related parent applications
• Related child applications
• Other family members
• PTAB proceedings (IPR / PGR trial metadata)

The fields returned are configurable via Actor input.

---

Input

patentOrPublicationNumbers or keywords or search_filters

You can provide one or more of:

• patentOrPublicationNumbers: a list of issued patent numbers and/or pre-grant/publication numbers
• keywords: one or more keyword queries that will be searched on Google Patents, e.g., (fraud detection OR theft detection) (bank)
• search_filters: inventor, assignee, patent office, and priority-date filters

The Actor resolves Google Patents search filters to patent/publication numbers first, then fetches the same metadata fields for the resulting documents.

You provide via the Form under the Input tab a list of patent/publication numbers and/or keywords and/or search filters and specify the fields you want to fetch for each result (title, inventor, assignee, priority date, expiration date, status):

Alternatively, you may provide JSON specifying the patent numbers, keywords, and fields:

{
    "patentOrPublicationNumbers": [
        "US3945785A",
        "US9439441B1",
        "US20250157242A1"
    ],
    "keywords": [
        "(fraud detection OR theft detection) (bank)"
    ],
    "inventor": "Sharad Deepak Sambhwani",
    "assignee": "Politechnika Warszawska",
    "office": "WO",
    "priorityDateFrom": "2021-01-01",
    "priorityDateTo": "2026-01-01",
    "maxKeywordResults": 25,
    "fields": [
        "patent_no",
        "publication_number",
        "application_number",
        "status",
        "priority_date",
        "publication_date",
        "exp_date",
        "abstract",
        "title",
        "inventor",
        "assignee",
        "claims",
        "parent_apps",
        "child_apps",
        "other_family_members",
        "ptab_proceedings"
    ],
    "delaySeconds": 0.5,
    "timeoutSeconds": 20,
    "maxRetries": 2,
    "useApifyProxy": false
}

For Google Patents searches, the Actor derives how many search pages to visit from maxKeywordResults and then keeps at most that many resolved patent/publication numbers per search combination. Example output is shown below:

When claims is selected, the Actor returns an array of claim objects for each patent, with:

• number
• type (independent or dependent)
• text

When parent_apps or child_apps is selected, the Actor returns an array of related application objects with:

• application_number
• publication_number
• relation
• priority_date
• filing_date
• title
• url

The related-family fields use both identifiers where available:

• application_number
• publication_number

When other_family_members is selected, the Actor returns family applications from the Family Applications Before and Family Applications After sections, with the same fields plus:

• direction (before or after)

The identifier fields behave as follows:

• patent_no is the issued patent identifier when one can be determined
• publication_number is the publication identifier shown on the Google Patents page when one is available (for example US20250157242A1 or US12361740B2)

When these fields are selected, the Actor keeps them separate:

• for a granted patent page such as US9439441B1, patent_no is populated with the granted identifier
• for a pre-grant publication such as US20250157242A1, publication_number is populated with that publication identifier
• if Google Patents exposes both versions, the Actor fills both fields in the same output row

This means a publication-number input can still produce a populated patent_no when Google Patents exposes the granted counterpart, and a patent-number input can still produce a populated publication_number when Google Patents exposes the publication counterpart.

When ptab_proceedings is selected, the Actor looks up matching IPR and PGR trial numbers from a PTAB proceedings database using the derived issued patent number. This means:

• granted patent identifiers such as US9439441B1 can be matched directly
• pre-grant publications such as US20250157242A1 can be matched when Google Patents exposes a granted counterpart in Other versions
• if no issued patent number can be determined, ptab_proceedings is returned as an empty list

Each ptab_proceedings item includes:

• trial_number
• petitioner
• status
• last_modified_datetime