All Collections
Usage instructions
Searching
Boolean Searching and Filtering of AI Searches and Collections
Boolean Searching and Filtering of AI Searches and Collections

Quick guide to IPRally Boolean search and filtering capabilities.

Sakari Arvela avatar
Written by Sakari Arvela
Updated over a week ago

Note: this document refers to the new BETA Boolean filters launched on June 30, 2023, and last updated on September 12, 2023. The old filters (Pre-May 2023) are deprecated and are/will be automatically converted to the new Boolean filters.

Background

Besides advanced Graph AI search capabilities, IPRally provides the capability:

  • to filter Graph AI search results and Collections using Boolean methods;

  • to search using Boolean methods.

This guide describes the syntax, gives examples and lists known limitations of the Boolean filters.

Syntax cheatsheet

Category

Operator/code

Description

Example(s)

Boolean operators

AND

Boolean AND

car AND battery

OR

Boolean OR

motor OR engine

NOT

Boolean NOT

motor NOT combustion

,

same as AND

car, battery

Proximity operators

Wn

Within n words of each other

filter W2 signal

NEAR

Within 5 words of each other

filter NEAR signal

NOTWn

Not within n words of each other

filter NOTW5 water

Wildcards

*

Unlimited left or right hand truncation

autom*

*correction

"multi* device"

?

Exactly one character

analy?e

"portable analy?er"

#

Zero or one characters

auto#correction

Note: at least 3 consecutive characters needed in the word when using a wildcard

Phrases

"..." OR ...

Multi-word phrase

"wind power plant"

wind power plant

"fib* board"

Grouping / nesting

(...)

(car OR automobile) AND (window AND (rear OR front))

Field codes

Number fields

PN:

Publication number

(auto-truncated)

PN:US90000000

AN:

Application number

(auto-truncated)

AN:EP14735465

Name fields

PA:

Applicant or original assignee

PA:microsoft

PA_CUR:

Current assignee

PA_CUR:(microsoft OR amazon)

PA_ALL:

Any applicant or assignee

PA_ALL:(nokia NOT (netw* OR solut*))

INV:

Inventor

INV:(kleuskens)

INV:("sarah kleuskens")

INV:("kleuskens sarah")

INV:("kleuskens, sarah")

Date fields

PD:

Publication date

PD:[2022-01-01 to *]

APD:

Application date

AD:[2020-02-01 to 2022-01-13]

PRD:

Priority date

PRD:[* to 2004-12-01]

Class fields

CPC:

CPC Class

(auto-truncated)

CPC:(A61F13 OR A61F20)

IPC:

IPC Class

(auto-truncated)

IPC:(A63B AND A61H3/00)

Text fields

TACD:

Any text

TACD:(car OR automobile) AND (window NEAR (rear OR front))

TAC:

Title, abstract or claims

TA:

Title or abstract

T:

Title

A:

Abstract

C:

Claims

D:

Description

Other fields

LS:

Legal status

LS:(ACTIVE OR UNKNOWN)

other values: INACTIVE

PBT:

Publication type

PBT:GRANT

other values:

APPLICATION

PBC:

Publication country

PBC:(US OR EP)

Stemming

Supported for T/A/C/D/TA/TAC/TACD fields

TACD:fisher

with stemming on finds

fish, fishing

Tags

Supported

Combining Custom Technology Tag queries to Boolean searching

Fielded searching/filtering

Supported

Fields combined by AND operator by default

Advanced searching/filtering

Supported

Allows complex queries

(PA:(daimler* OR mercedes*)) AND (PD:[2020-01-01 to *]) AND (TACD:((car OR autom*) AND (window NEAR (rear OR front)))) AND (TA:vehicle) AND (LS:(ACTIVE OR UNKNOWN)) AND (PBC:(US OR EP))

Case insensitivity

Supported

All text fields are case insensitive

Motor = motor = MOTOR

Relevance scoring / ranking

Supported

The results are scored and ranked by Boolean relevance, according to a nativeRank /
โ€‹nativeFieldMatch algorithm

Boolean scoring

If you use Boolean text fields (T/A/C/D/TA/TAC/TACD), you will get a Boolean score ("Bn" score) in addition to the AI score. The best match will get a Bn score of 100 and the other matches a lower score, using an advances text scoring algorithm. For forming the score, the prioritisation of the fields is T > A > C > D, but there are also other affecting factors, like the term count, term proximity and so on.

Limitations (as of 2023-09-12)

Limitation

Explanation

Plan

Number of supported documents for search result exports

The number of supported documents in exporting documents is limited to the number of paginated (loaded) documents in the user interface (and max 10k)

We're adding better result pagination and support for

  • exporting of larger result sets

  • adding larger result sets to Collection

Search speed with Wildcards

Search sometimes takes too long when there are Wildcards in the query

Investigating and improving during H2/2023

Combined Proximity operators

Combined expressions like (a NEAR b) NEAR c are not supported

Support is coming later

Wildcards need at least three consecutive characters in some part of the word

mot?r is a valid form

mo?or in not a valid form

Let us know if this is too limiting in practice!

View customization

The FIELDED search/filtering view is not yet customizable (all fields shown always)

Customization possibility coming later.

Fielded search/filter vs. Advanced search/filter

Fields in FIELDED SEARCH/FILTER are combined in the background using AND operator. The following corresponds to (PN:US9) AND (PA:micro*):

Advanced search allows search for (PN:US9) OR (PA:micro*) too:

Above the SEARCH PATENTS button, one can easily see and copy the corresponding Full boolean query to clipboard.

Search field validation

The fielded search aims to capture incorrect/unsupported syntax automatically when the field is unfocused.

Note: not all incorrect syntaxes are captured yet, whereby you may get an error or empty result list when pressing the SEARCH / FILTER button.

Filtering of AI search results

Collection filtering

Did this answer your question?