Thomson Innovation Search Fundamentals

This information will help you craft queries that produce relevant and manageable result sets. These fundamentals apply to patent, business, and literature searches.

Detailed information on the forms, fields, and strategies specific to patent, business, or literature searching can be found in:

Search Operators Summarized

Operator	Patents	Business	Literature
AND	Both terms must be present
OR	One term or the other must be present
NOT	Term must be excluded
ADJ	Terms should be next to each other and in order specified
ADJn	Terms should be within n words of each other and in the order specified
NEAR	Terms should be next to each other and in any order
NEARn	Terms should be within n words of each other and in any order
SAME	Terms must be in the same paragraph, in any order
=	Equal to (use with text as well as dates and amounts)
<>	Not equal to
>	Greater than
>=	Greater than or equal to
<	Less than
<=	Less than or equal to
( )	Create nests to define order of operation
?	Stands for one character, can use multiples, can use within a term	Multiple question marks in sequence following a stem stand for an equal number of characters Within a term, stands for one character, can be used in multiples	Stands for one character, can use multiples, can use within a term
*	Stands for zero to many characters, can use within a term
*n	Stands for zero to n characters, can use within a term
{d}	Stands for a digit (0-9); can use multiple, {d}{d} Note: Only available with Expert style patent search	Not available	Not available
{c}	Stands for a consonant; can use multiple, {c}{c} Note: Only available with Expert style patent search	Not available	Not available
{v}	Stands for a vowel; can use multiple, {v}{v} Note: Only available with Expert style patent search	Not available	Not available
{a}	Stands for a letter (A-Z); can use multiple, {a}{a} Note: Only available with Expert style patent search	Not available	Not available

Search Operators in Detail

A search operator is a word or symbol used for expressing a function that should be performed with the specified keywords.

In the following examples, the search operators are shown as all uppercase and, when you use the search form tools to construct your queries, they will also be shown as upper case. This is, however, not a requirement and is done only to visually differentiate the operators from the surrounding keywords.

On search forms, you can choose operators from drop downs or add them using convenient buttons. You can also enter them yourself when constructing queries from scratch.

Operator

Collection

Use and Examples

AND

All

Searches for two terms that must both be present.

This operator places no condition on where the terms are found in relation to one another; the terms simply have to appear somewhere in the same document.

Examples:

printer AND scanner

(Hewlett-Packard OR Lexmark) AND (printer AND scanner)

All

Searches for two terms when one term or the other must be present.

This operator places no condition on where the terms are found in relation to one another; one or both terms simply have to appear somewhere in the same document.

Examples:

printer OR scanner

(Hewlett-Packard OR Lexmark) AND (printer OR scanner)

NOT

All

Specifies a term that should be excluded.

This operator retrieves documents that contain the first search term you specify but not the second term.

Examples:

printer NOT scanner

(Hewlett-Packard OR Lexmark) AND (printer NOT scanner)

Note: There are some special restrictions that apply to the use of NOT, see NOT Operator Restrictions.

ADJ
ADJn

Patent

Business

(see Literature following)

Searches for terms within one word of each other (adjacent), in the order specified.

When followed by the numeric qualified n, searches for terms in the order specified and within n terms of each other. The value n specifies the number of searchable terms (minus one) allowed between search terms.

Examples:

inkjet ADJ printer

wave ADJ3 energy

Note: In business searches, this is the equivalent of the Dialog (W) and (nW) operators.

ADJ
ADJn

Literature

In literature searches, ADJ and ADJn work like SAME and they look for terms all in the same sentence, in any order.

NEAR
NEARn

Patent

Business (see Literature following)

Searches for records containing the specified terms in any order.

When followed by the numeric qualified n, searches for records containing the specified terms within n number of words of each other in any order. The value n specifies the number of searchable terms (minus one) allowed between search terms.

Examples:

GMC NEAR finance

Note: In business searches, this is the equivalent of the Dialog (N) and (nN) operators.

NEAR
NEARn

Literature

In literature searches, NEAR and Nearn work like SAME they look for terms all in the same sentence, in any order.

SAME

Patent
Business
(see Literature following)

Searches for terms all in the same paragraph, in any order.

Example:

humayan NEAR optical

Note: In business searches, this is the equivalent of the Dialog (S) operator.

SAME

Literature

Searches for terms all in the same sentence, in any order.

All

Equal to. Can be used for dates, numeric terms, and text with patent expert search style.

Examples:

pd = (20040527)

Field name	Field tag	Operator	Search term
Publication Date	pd	=	20040527

rv = ($4M)

Field name	Field tag	Operator	Search term
Revenue	rv	=	$4M

in = (dean kamen)

Field name	Field tag	Operator	Search term
Inventor	in	=	dean kamen

NOTE: When searching a date-range, the dates must be specified in ascending order, e.g., >=20010101 <=20011231.

All

Not equal to. Can be used for dates and other numeric terms.

Example:

ay <>= (2004)

Field name	Field tag	Operator	Search term
Application Year	ay	<>	2004

All

Greater than. Can be used for dates and other numeric terms.

Examples:

rad > (20040527)

Field name	Field tag	Operator	Search term
Related Application Date	rad	>	20040527

rv > ($4M)

Field name	Field tag	Operator	Search term
Revenue	rv	>	$4M

NOTE: When searching a date-range, the dates must be specified in ascending order, e.g., >=20010101 <=20011231.

All

Greater than or equal to. Can be used for dates and other numeric terms.

Examples:

pd >= (20040527)

Field name	Field tag	Operator	Search term
Publication Date	pd	>=	20040527

rv >= ($4M)

Field name	Field tag	Operator	Search term
Revenue	rv	>=	$4M

NOTE: When searching a date-range, the dates must be specified in ascending order, e.g., >=20010101 <=20011231.

All

Less than. Can be used for dates and other numeric terms.

Examples:

pcpd < (20040527)

Field name	Field tag	Operator	Search term
PCT Publication Date	pcpd	<	20040527

rv < ($4M)

Field name	Field tag	Operator	Search term
Revenue	rv	<	$4M

NOTE: When searching a date-range, the dates must be specified in ascending order, e.g., >=20010101 <=20011231.

All

Less than or equal to. Can be used for dates and other numeric terms.

Examples:

pcpd <= (20040527)

Field name	Field tag	Operator	Search term
PCT Publication Date	pcpd	<=	20040527

rv <= ($4M)

Field name	Field tag	Operator	Search term
Revenue	rv	<=	$4M

NOTE: When searching a date-range, the dates must be specified in ascending order, e.g., >=20010101 <=20011231.

()

All

Parentheses create nests which help define the order of operation.

Nesting directs the search engine to process your query in an exact order, avoiding misunderstandings. Search instructions within parentheses are always processed first.

Examples:

(line OR string) AND trimmer

driving AND (protection OR helmet)

NOTE: Order of operation is also determined by the operator. See more about operator precedence.

Patent

Literature

(see Business following)

The question mark wildcard represents one character.

Use one or more question marks to stand for a specific number of characters in your search term. The question mark can be used within a word.

Examples:

carbo?
Results include carbon, carbox, and carboy

carbo??
Results include carbons, carboxy, and carbony

car?on
Results include carbon and carton

???oxide
Results include monoxide, peroxide, hexoxide, and ethoxide

NOTE 1: A single question mark easily compensates for differences between US and British spelling. For example: sterili?e or t?re.

NOTE 2: When a wildcard is used in a search term, stemming is disabled for that term. See more about stemming.

Business

Multiple question marks in sequence following a stem stand for an equal number of characters

Within a term, stands for one character, can be used in multiples

*
*n

All

The asterisk wildcard represents zero or an unlimited number of characters. The asterisk can also be used within a word.

If the asterisk is followed by a numeric quantifier (*n), the quantifier indicates the exact number of characters desired.

Examples:

carbo*
Results include carbo, carbon, and carboxypeptidases

carbo*ate
Results include carbonate, carbohydrate, and carboxydiketonate

carbo*2
Results include carbons, carboxy, carbody, and carbony

NOTE: When a wildcard is used in a search term, stemming is disabled for that term. See more about stemming.

Default Operators

In your search preferences, you can change the defaults for both between fields and within fields. For between fields, you can choose to use AND, OR, or NOT as your default. For within fields, you can choose to use AND, OR, or ADJ as your default.

NOT Operator Restrictions

These restrictions are in addition to the restrictions concerning NOT as a search term, see Stopwords and Reserved Words for restrictions on using NOT as a search term.

* and ? Operator Considerations

Overview

Use of the * and ? operators (often referred to as wildcards or truncation operators) is limited by the scope of your search. Scope influencing factors include the number and size of the collections chosen, the fields being searched, and the ubiquity of the search criteria.

Example 1:

In this example, the search string Qu* contained too few regular expressions for the scope of a search that encompasses all text fields. The search string Qu* does return results when the scope of the search is limited to a field like Inventor that is much narrower in scope than all text fields.

But, if you add just one more regular expression (in this case, the letter e), the scope of the search is narrowed enough that results can be retuned even when all text fields are searched, as shown following.

Example 2:

In this example, P* contains too few regular expressions to create an Inventor search in either the US Granted or French Applications collections.

Following you can see that adding one regular expression (in this case, the letter e) will not impact the scope of the search enough to make it viable in the US Granted collection, but will give you results in the smaller French Applications collection.

You need to add a third regular expression (in this case, the letter t) to get results from a search of the US Granted collection, as shown following.

Example 3:

The search above is asking for every term that starts with P and ends with T in every text field and the scope is much too broad to return results.

This search (above) is asking for every 6-character term that beings with P and ends with T in every text field and the scope is still too broad.

This last search is asking for every 5-character term that beings with P and ends with T in every text field and the scope is still manageable. Results include the terms print, point, plant, and paint.

Best Practice

While the combinations of collections, fields, and search criteria are not infinite, whether or not your query can be processed is actually determined by internal system algorithms. This makes it impractical for a user to try to determine in advance whether or not their * or ? search will return results. The best practice is to limit your use of these operators as much as possible. When a search is too broad to be processed, an error message is returned.

Operator Precedence

Query expressions are read using specific rules of operator precedence. This means that certain operators are processed before others. While query expressions are read from left to right, some operators are processed before others and impacts the way the search engine treats your query.

This following example shows how the precedence rules can impact the manner in which the search engine processes your query.

If you want to search for patents about feline disease or ferret disease, and you enter

because AND is treated before OR, the search engine will interpret your query to mean this

and your result set will include records with feline and disease or records with ferret that may or may not include the term disease.

This is a better way to construct the query

Now all records in your result set will contain the word disease and either the word feline or the word ferret.

Case Sensitivity

Stopwords

Stopwords have been implemented for business searching -- but not for patent or literature searching. Stopwords are not permitted as query terms.

Reserved Words

Certain words are reserved for use as operators, and, if your search string includes a reserved word, it will be interpreted as an operator. Reserved words can be searched in the database, but must be entered within double quotes to distinguish them from operators.

Reserved words for patent and literature searching are: AND, NOT, OR, SAME, WITH, and NEAR.

Reserved words for business searching are: NOT, OR, SAME, and NEAR. AND and WITH are not included in this list because, in business searching, AND and WITH are stopwords and cannot be used at all. See Stopwords for more information.

To search for a reserved word per se, type the word in double quotes: e.g., "near". To search for the phrase 'near field', you should type the following: "NEAR" ADJ FIELD.

Hyphenated Words

In order to effectively search for hyphenated terms, you need to understand how they are indexed. In the patent database, hyphens are not indexed; they are treated as if they were a space. If a hyphenated term is used in a search query, it is converted to an adjacency syntax. Note that this holds true even when the hyphen is enclosed in quotation marks.

So, for example "computer-related" is equivalent to "computer related" and "semi-conductor" is equivalent to "semi conductor". To search for the former, you would type

where the adjacency operator (ADJ) will account for both the hyphenated and the non-hyphenated forms and "semiconductor" will find those instances where the word is spelled without hyphen or space.

Exact Words vs. Stemming

Stemming extends a search to cover different variations of a word. This means when you search on a word like prime, your result set will include words that share a root, or stem, with the word you searched. So, for prime, your result set will include words like primed, priming, primaries, and primates.

Stemming is a linguistic process and your results will include linguistic expansions of the stem word. Use wildcards for a result set that includes all expansions of a stem or word. Stemming is not applied to any search term that includes a wildcard.

By default, stemming is Off. Change the default to On from your Search Preferences screen. See more information on search preferences and defaults.

Date Format

Dates can be searched in date fields in any of three formats (YYYYMMDD, YYYY-MM-DD, or YYYY/MM/DD). In query Previewer, Search History, and Saved Queries, the data format will be displayed as YYYYMMDD, regardless of the input format. All dates in explicit date fields can be searched as full date or as year, and additionally can be searched as date ranges (>20070101, >=20070101, >=20070101 <=20070801).

In fields that include a date associated with other information, the date format used must be YYYYMMDD and additionally, range searching is not permitted. These fields include the following:

Field	Search Criteria	Collection	Result
Text Fields	P*T	US Granted	Scope of the search is too broad

Search Term	Result Set Includes
prime	prime, priming, primed, primates, primaries, and other words with the same stem
carbon	carbon, carbons, carbonate, carbonates, carbonated, and other words with the same stem

Inventor	PE*	US Granted	Scope of the search is too broad
Inventor	PE*	French Applications	Appropriate results returned