 |
|
   |
 |
 |
|
|
|
|
|
|
|
|
|
|
See Field Help: Definitions and Searching in the Delphion Help Center for field-specific definitions and abbreviations. See Using Corporate Tree in the Delphion Help Center for information on using Corporate Tree for Assignee searches. Overview This information will help you craft your queries to produce more relevant and manageable result sets. Queries that are carefully and purposely created let you spend less time looking for information and more time acting on it. [back to top] Entering Queries on the Delphion Website Use the Advanced search page to enter your own queries on the Delphion website. The Advanced Text Search section of the page (at the top) allows you to specify which collections you want searched and how many queries should be displayed per page in your result set. There are also drop down boxes that allow you to specify a date range. When you receive your result set, the top of the result set page will show you how Delphion has reformatted your query to make it flow quickly through the search engine. If your query does not return the expected results, you may want to study the reformatted query to see if there are clues to what might be the problem. Also at the top of the result set page, you will find a text entry box in which the reformatted query is repeated. Use this text entry box to change your query — or enter a new one. Immediately following the text entry box is confirmation of the collections searched to produce this result set. [back to top] Complex Queries In general a complex query may contain any of the following:
[back to top] Boolean Operators Use Boolean operators to indicate which keywords you want included in, or excluded from, your search result set.
Example 1, the AND operator: To search for all patents with Hewlett-Packard as the Patent Applicant/Assignee (PA) and both the words printer and scanner in the Title (TI), use the following query: (Hewlett-Packard <in> PA) AND (printer AND scanner) <in> TINote that printer and scanner as well as the Boolean operator AND are enclosed by parentheses. The parentheses indicate that both the words printer and scanner should be in the TI field. If the parentheses were omitted, the result set would include patents with the word printer shown in any field in the patent record and with scanner in the TI field. Example 2, the OR operator: To search for all patents with Hewlett-Packard as the Patent Applicant/Assignee (PA) and with either the word printer or the word scanner in the Title (TI), use the following query: (Hewlett-Packard <in> PA) AND (printer OR scanner) <in> TINote that, like the example above, the words printer and scanner as well as the Boolean operator are enclosed by parentheses. The parentheses indicate that both the words printer and scanner should be considered when the TI field is searched. In this case, the Boolean operator is OR, so the patent qualifies if either (or both) of those words is found in the TI field. Example 3, the NOT operator: To search for all patents with Hewlett-Packard as the Applicant/Assignee (PA) and with the word printer but not the word scanner in the Title (TI), use the following query: (Hewlett-Packard <in> PA) AND (printer NOT scanner) <in> TILike the other examples, the words printer and scanner as well as the Boolean operator are enclosed by parentheses so that both of the words are considered when the TI field is searched. In this case, the Boolean operator is NOT, so the patent will qualify for the result set only if the word printer is present and the word scanner is not present. Example 4, stringing NOTs: You can string or use multiple occurrences of the NOT operator to exclude several terms from your query. To include the word printer but exclude these common variations of the phrase ink jet from your results, use the following query: (Hewlett-Packard <in> PA) AND (printer NOT "ink-jet" NOT "inkjet" NOT "ink jet") <in> TIStringing AND NOTs produces exactly the same results but requires more key strokes. However, if you prefer, you could construct your query like this: (Hewlett-Packard <in> PA) AND (printer AND NOT "ink-jet" AND NOT "inkjet" AND NOT "ink jet") <in> TI[back to top] Using the <in> Operator <in> is a proximity operator that helps you select documents which contain your keywords in specified fields in the patent record. Proximity operators are always enclosed in angle brackets (less than and greater than symbols). They are not case sensitive so you may use upper or lower case. When using the <in> operator, you need to specify the field you want to have searched. You can use the accepted full field name or an accepted abbreviation for the field name. The following two queries return exactly the same result set: Query with full field name:monoclonal <in> title Query with field name abbreviation: monoclonal <in> TI Field name abbreviations are usually shown all upper case but they are not case sensitive either. The following returns the same result set as the two preceding queries: Query with field name abbreviation in lower case:monoclonal <in> ti Using natural language for the field name will not work. The following query will return a message advising that your query is improperly formed: Field name improperly formed:monoclonal <in> the patent title NOTE: The <in> operator does not work with date fields. [back to top] Searching for Phrases By default, the search engine performs phrase searching. This means two or more words entered without commas or other operators will be treated as a unit — a phrase — when the search is performed. So, a search for optical fiber will return a result set that includes optical fiber, optical fibers, and optic fiber. The words will always be adjacent and always be in the specified order. Variations of the words will appear because stemming is also, by default, turned on. If you want your search to include the words optical and fiber in any order and not necessarily next to each other, then construct your query in a way that turns phrase searching off.
See also: Searching for Exact Terms vs. Stemming [back to top] Date Range Searching Date Range Searching allows you to define or restrict the dates of your search. Use the name of a date field followed by one or more of the mathematical symbols for less than, equal to, or greater than. Then specify the dates you want searched. For the best results, always use the ISO date format: YYYY-MM-DD.
NOTE: The less than, equal to, and greater than symbols only work with date fields, they cannot be used with text fields. Example 1, dates earlier than: To search for all patents with Publication Date (PD) before (less than) 1991-01-01 with Canon as the Patent Applicant/Assignee (PA) with the word image in the Title (TI), use the following query: (Canon <in> PA) AND (image <in> TI) AND (PD<1991-01-01)Notice that each of the three main elements of this query are enclosed by parentheses to ensure that each element is considered independently when patent fields are searched. In this case, the Boolean operator AND is used between each of the query elements to indicate that positive results in all three elements of the query are required for a patent to appear in your result set. Example 2, dates earlier than or equal to: To use the same basic query as the one shown in Example 1, but this time include patents actually issued on 1991-01-01, use both the less than and equal to symbols as shown in this query: (Canon <in> PA) AND (image <in> TI) AND (PD<=1991-01-01)The result set for this query will contain all the patents shown in the result set from the Example 1 query and will also contain the patents with Publication Date (PD) actually equal to 1991-01-01. Example 3, dates between two dates: To search for patents with Publication Dates (PD) between 1988-12-31 and 1991-01-01, construct your query like this: (Canon <in> PA) AND (image <in> TI) AND (PD>1988-12-31 AND PD<1991-01-01)1989-01-01 is the first date that is greater than 1988-12-31 and 1990-12-31 is the first date less than 1991-01-01. So this result set will include patents issued on 1989-01-01 and patents issued on 1990-12-31 — as well as all the patents issued in between (as long as they meet the rest of the specified criteria). Example 4, dates between two dates, including the start and stop dates: To search for patents with Publication Dates (PD) between 1988-12-31 and 1991-01-01, but including those two dates, construct your query like this: (Canon <in> PA) AND (image <in> TI) AND (PD>=1988-12-31 AND PD<=1991-01-01)1988-12-31 will be the first date included in your result set. 1991-01-01 will be the last date included in your result set. You will also get all of the dates in between. NOTE: When searching large or multiple collections (which return more than 500 records for your query) you will not be able to see all of the patents that fall into your desired date range. If this is a problem, you should rewrite and target your query to return a result set that is under 500 records and therefore can be viewed easily and manipulated accurately — and using date ranges is usually the easiest and quickest way to do this. [back to top] Searching for Exact Words vs. Stemming The Delphion search engine automatically performs stemming. This means when you enter a word such as 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 only applies to words with four or more characters. If you do not want stemming used for your search, you need to specify in your query that only the exact keyword should be searched. Do this by enclosing your keyword(s) in double quotes.
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 (right-hand) expansions of a stem or word. See also: Searching for Phrases and Wildcards [back to top] Nesting and Order of Operation Parentheses are used to create nests which define the order of operation. Nesting directs the search engine to process your query in an exact order, avoiding misunderstandings. The two queries below will retrieve radically different result sets. As in algebra, what appears inside parentheses is processed first.
[back to top] Order of Precedence Query expressions are read using specific precedence rules. This means that certain operators are processed before others. While query expressions are read from left to right, some are processed before others and this can impact the way the search engine interprets your query. The following shows the order or precedence in which operators are processed:
Example 1, AND before OR: If you want to search for patents about feline disease or ferret disease, and you enter ferret OR feline AND diseasebecause AND is treated before OR, the search engine will interpret your query to mean this ferret OR (feline AND disease)and your result set will include records with feline and disease or records with ferret that may or may not include the term disease. So here is a better way to construct the query (ferret OR feline) AND diseaseNow all records in your result set will contain the word disease and either the word feline or the word ferret. Example 2, <in> before AND: If you want to search for patents with the words feline and disease in the Title, and you enter feline AND disease <in> titlebecause <in> is treated before AND, the search engine will interpret your query to mean this feline AND (disease <in> title)and your result set will include records with disease in the Title and feline in any searchable field. So here is a better way to construct the query (feline AND disease) <in> titleNow all records in your result set will contain both the words feline and disease in the Title. [back to top] Relevancy Scoring Each record in your Result Set has a score (shown on the far right of the Result Set page). The score is an indication of that record's relevancy to your query. The score is based on algorithms built into our search engine which factor in the density of the search terms found in the document. Density is calculated in proportion to overall document text — because of this, a longer document that contains more occurrences of a word can score lower than a shorter document that contains fewer occurrences. And, when proximity operators are used, the nearness of terms within the document is also a factor. The exact details of the scoring algorithm are proprietary. This is the default scoring mechanism. The relevancy score is most useful when searching for key words in text. It’s less useful when you’re searching for one instance of a term in specific fields, such as in IPC searching or assignee searching. If you prefer not to use the relevancy score, use the "Use Relevancy Score?" toggle on the Current Results tab of your Result Set to turn it off. You can also easily turn it off by inserting the <yesno> operator before your query like this: <yesno> (carbon <in> ti)When relevancy scoring is on, the Result Set is sorted in relevancy order. When scoring is off, the Result Set is returned in date order. You can easily re-sort your Result Set by clicking the desired column header on the Result Set page. [back to top] <near> <near/n> and <order> Operators The <near> and <near/n> proximity operators allow you to retrieve documents which are relevance-ranked based on the closeness (proximity) of the specified words. <near> selects documents containing your specified search terms. Delphion's search engine scores a 0 for documents where the search terms are not within 1024 words of each other. The closer the search terms are within a document, the higher the document’s relevancy score. <near/n> selects documents containing two or more search terms within a specified number of words of each other. n can be an integer up to 1024. Once again, the closer the search terms are within a document, the higher the document’s score. Neither the <near> or the <near/n> operators specify the order of the terms. To specify the order of the terms, use <order> preceding the <near> or <near/n> operators. Note: You cannot use a NOT command in conjunction with a proximity operator. Example 1, the <near> operator: The following queries demonstrate the syntax for the <near> operator: coffee <near> filter (This Result Set will include all records with the word coffee near the word filter.) (coffee <near> filter) <in> description (This Result Set will include all records with the word coffee near the word filter in the Description field.) (coffee <near> filter) <in> (abstract,claims) (This Result Set will include all records with the word coffee near the word filter in the Abstract or Claims fields.) ((dog OR cat) <near> carrier) <in> description (This Result Set will include all records with the word dog or the word cat near the word carrier in the Description field.)It is important to remember that the closer the search terms are to each other, the higher the relevancy score of the document. Example 2, the <near/n> operator: The following queries demonstrate the syntax for the <near/n> operator as well as the difference in Result Sets when the variable is changed: (inorganic <near/10> solvents) <in> claims (over 1,000 records found for collection searched)Remember that the closer the search terms are to each other, the higher the relevancy score of the document. Example 3, the <order> operator used in conjunction with <near> and <near/n>: The following queries demonstrate the syntax for using the <order> operator to specify the order in which the target words should appear to meet the criteria for your query when using the <near> and <near/n> operators. ((dog OR cat OR pet) <order><near> (cage OR carrier)) <in> descriptionFor this query, records with the words dog, cat or pet near the words cage or carrier will be found — but only those instances in which dog, cat or pet precede the word cage or carrier. If the <order> operator had not been used, then the Result Set would also include instances in which cage or carrier appeared before dog, cat or pet. (((wire OR mesh OR screen) <order><near> (dog OR cat OR pet)) <order><near> (cage OR carrier)) <in> abstractThe Result Set for this query contains records with the word wire, mesh or screen near but preceding the word dog, cat or pet which is near but precedes the word cage or carrier. Following is an abstract from a patent in the Result Set for this query: A collapsible pet enclosure that is fashioned to fit in standard windows and has removable material coverings andscreen panels on all sides. Domestic animals can use the enclosure to enjoy the out doors with out the owners fearing for their safety or health, as is the case in letting pets out of doors on the ground level. There are detachable wheels as part of the unit that convert the enclosure to a mobile pet carrier, that one may roll instead of carry. The <near/n> operator works in much the same way — but allows you to specify the maximum distance between your target words and still have the words appear in a preferred order. Following is an example of using <order> with <near/n>: (((inventory OR component OR parts) <order><near/10> (management OR tracking OR location) <order><near/15> ("bar code" OR hand-held OR wireless)) <in> abstractThis Result Set for this query will include all records with management or tracking or location within 10 words of inventory or component or parts and then, within 15 words of that word combination, the word(s) "bar code" or hand-held or wireless must appear. When this search is done with the <near/n> operator, a Result Set shows 17 patents — done with just the <near> operator, the Result Set had over 100 patents. So you can see how the <near/n> operator can help refine your search. Following is an abstract from a patent in the Result Set for this query: A method ofinventory management is described. Upon activation of a button on a wireless device, the wireless device having a light source and a transceiver with a unique media address corresponding to a unique product, the device broadcasts a first signal including an order command and the unique media address by the transceiver via a wireless medium. A central controller then receives the first signal, identifies the unique media address included in the first signal, and using a database, identifies the unique product associated with the unique media address. [back to top] Thesaurus Operator The THESAURUS operator searches an index of terms and finds synonyms for the key term in your query. The following query asks the search engine to look for the word bow in the Title (TI) field. bow <in> TIThe Result Set includes patents for an archery bow, bow tuning equipment, a gift wrapping bow, a bow assembly for a ski lift, a key bow, a ladder for a boat bow, a bow rake, and a violin bow cover. All patent records have the word bow in the Title. For the collection searched, the Result Set was over 1,000. This query uses the <THESAURUS> operator to ask for results that include synonyms for the word bow in the Title. <THESAURUS> bow <in> TIThis Result Set includes patents with the word bow and patents with synonyms for the word bow in the Title. Synonyms found include: bend, bending, turn, turnable, yield, yielding, curved, curving, round, and crook. For the collection searched, the Result Set was over 10,000. NOTE: The thesaurus includes mostly common words and may not be helpful for technical or scientific terms. [back to top] Wildcards A wildcard is a symbol that stands for one or more unspecified number of characters in a query. Delphion supports the use of wildcards in the right-hand position, after a specific character string (known as right-hand truncation), in queries for all collections. This means that you can use both the asterisk and the question mark wildcards to the right-hand side of a search term. This allows you to retrieve words that begin with a specific character string but have one or more unknown characters at the end. For left-hand wildcards (known as left-hand truncation), using both the asterisk and the question mark are supported for both the German Applications and German Granted collections. For all other collections, only the question mark is supported for left-hand truncation. This allows you to retrieve words that have one or more unknown characters at the beginning, but end with a specific character string. The following wildcard symbols are valid:
Following are examples of queries using wildcards:
[back to top] Weighting Search Terms You can assign a weight to each search term in a query to indicate the term’s relative importance to your search. The weight assignment is shown as a number between 01 and 100, where 01 represents the lowest importance rating and 100 represents the highest. You would never use weighting with just one search term because you are weighting (or comparing) one term against another. For example, you are working on a toothbrush holding device and want to check for related patents. You are interested in patents on toothbrush holders but also patents on holders of any sort. On a relative scale, patents on toothbrush holders are twice as important to you as patents on other kinds of holders, so you weight them accordingly in your query. To construct this kind of query, enclose the weight in brackets and join the terms with OR, like this: ([100](toothbrush AND holder) OR [50]holder)When sorted by score, the Result Set for this query will give patents with both the terms toothbrush and holder a higher score than those with just the term holder. Restricting your query in terms of time range (e.g., to a single year) will help highlight the scoring differences. [back to top] <accrue> Operator The <accrue> operator selects documents containing at least one of the search terms you specify (it is somewhat like OR in that manner). <accrue> however ranks documents according to the number of times the search elements appear in a given document — the more occurrences, the higher the score. The <accrue> operator accrues, or adds up, the number of occurrences of all of the terms combined for scoring. Because of this, <accrue> is not valid with just one search term. Construct your query like one of these: toothbrush <accrue> holder[back to top] Relevancy Scoring, Weighting, and <accrue> Compared Weighting, relevancy scoring, and <accrue> are three search options that are somewhat related and often confused. If you ask how documents are scored, the brief answer is that they are scored by frequency of key words. There are however, different aspects of that scoring that you should be aware of as well as different ways in which you can influence the scoring. Relevancy score Relevancy scoring is based on the density of search terms in the retrieved documents. This is the default scoring mechanism. Weighting Weighting is something you request when you form your query. You tell the search engine how you want the qualifying documents weighted against each other. <accrue> The <accrue> operator selects documents that include at least one of the search elements you specify. The more search elements that are present, the higher the score will be. Results The following demonstrates how a changing the way a query is formed can impact relevancy scoring. While these queries are not completely parallel, they show how one patent can receive different scores in similar queries. Default relevancy scoring decided the scores for these two queries: (toothbrush OR holder) <in> AB = 97%Weighting of terms when the query was constructed produced this score: (([100](toothbrush AND holder) OR [50]holder) <in> AB) = 98%The <accrue> operator was used to produce this score: (<accrue> (toothbrush, holder)) <in> AB = 70%Actual occurrences of the search terms in the patent abstract:
Alternative Languages (EP and WO collections only) For EP and WO patents, the Title, Abstract, Claims, Description, and Text fields can be searched in German, French, and Spanish. Use the following search terms in your query:
Format your query in the following manner: aufzug <in> titledeNOTE 1: These queries search only those records that have data available in the specified language, within the specified collection. NOTE 2: English is the primary or default language for Delphion and for the Delphion Integrated View. This means that, when there is an English language version of field information, the English will display on the Integrated View. For example, your Result Set for the search aufzug NOTE 3: When searching the German collection, you do not need to use titlede, abstractde, claimsde, descriptionde, or textde because the primary language of the collection is German. [back to top] |
| ||
|
 |
Copyright © 1997-2013 Thomson Reuters |
| Subscriptions | Web Seminars | Privacy | Terms & Conditions | Site Map | Contact Us | Help | ||
