New workflow: Patent analysis

Analyzing patents is tedious. First, you need to write complex search queries, hoping they'll capture your topic's essence. Even then, you must check each result for relevance—no search query, no matter how well-crafted, can perfectly match your needs.

Once you've identified the truly relevant patents, you face another round of analysis: What technologies do they protect? What applications do they target? Answering these questions requires diving deep into each patent's claims and other content.

Spark's new patent analysis workflow automates these steps for you. Here's how it works:

Search query generation

Spark takes your topic and generates a patent search query. These queries follow a specific structure. A verification engine ensures the generated query meets all requirements—including proper Boolean logic, bracketing syntax, wildcards, and patent classes.

You can edit your query. When you do, Spark will regenerate the search query explanation to match your edited version.

Spark focuses on most recent patents

In its search, Spark focuses on the most recent patents based on their publication dates (rather than priority dates).

You can analyze between 10 and 100 patents. When analyzing 100 patents, the process takes approximately 10 minutes—during this time, you're free to do other tasks, just ensure you keep the browser tab with the patent analysis open.

Patent relevance assessment

Why

Even a patent that matches your search query may not be relevant to your topic—it's virtually impossible to create a query that captures exactly your topic and nothing else.

In other words, despite careful work on search queries, each result still needs to be checked for relevance.

Spark automates this for you.

How

When assessing relevance, Spark examines both the abstract and claims of each patent.

For each patent, Spark assesses whether a PHOSITA (person having ordinary skill in the art) would consider the patent relevant to the topic at hand.

Spark evaluates this using three main criteria:

• Technical field: Is the patent's technical field related to the topic area?
• Problem domain: Does the patent address challenges relevant to the topic?
• Solution approach: Could the proposed solution be applicable to the topic, even if only partially described?

While additional criteria exist, these three are often considered the most important.

In order for a patent to be considered "relevant", all three criteria have to be met.

Spark extracts technologies and applications from patents

After determining which patents are relevant, Spark analyzes their key technologies and applications (skipping non-relevant patents). This analysis considers the claims, abstract, and when available, the full patent description.

Within the claims section, Spark first determines what the the independent claims are and examines those. Then, it goes through the dependent claims, to identify more specific technologies and applications.

You can view the results either grouped by technologies or applications...

...or by patents:

All analyses are stored in your personal data vault

When an analysis is finished, Spark automatically saves the results to your personal data vault. You—and only you—can access these results anytime through "My Contents" → "Patent Analyses".

Download analyses as Word documents

You can download analysis results as Word files directly from the "Patent Analysis" workflow, or later from the stored results (under “My Contents”). This makes it easy to share your work with others.