Search Help

• How does Funnelback search work?
• What is a word?
• How are simple queries processed?
• Search Fields
• Accented Characters
• Query Language Operators
• Example Queries
• Search Results

How does Funnelback search work?

Funnelback processes a query consisting of one or more words or a phrase (specific series of words), and tries to find documents containing all of the words. However, Funnelback will also find "partially matching" documents, which contain only some of the words in the query.

What is a word?

Funnelback considers any unbroken sequence of letters and/or numbers, such as Jones, a, 1999, or ENGN3410 as a word, and it does not distinguish between upper and lower case. Except for rare examples such as IT, case insensitivity produces the best search results.

Compound words such as non-adhesive, O'Toole, www.nrc-cnrc.gc.ca, or reports/1999 also act like single words when used in a query, although internally they are processed as phrases.

Unusually formed words may include punctuation marks, hyphens or other non-alphanumeric characters. For example, the chemical compound 3,3'-dipyrrolyl may appear to the search engine as more than one word. To find this whole word and not parts of it, enclose the word in quotation marks in the search query. Funnelback will then find an exact match rather than returning a large number of partially matching results.

Funnelback search limits words to a maximum of 20 characters in full-text fields (such as the Anywhere field) and 18 characters in metadata fields. That is, users can enter words longer than 18 or 20 characters but Funnelback uses only the first 18 or 20 characters of the search term to perform the search.

Stopwords such as the, a, and of occur frequently in documents and usually slow down query processing while contributing little to the quality of results. However, in some cases, e.g. the phrase "to be or not to be" they can be important. Accordingly, Funnelback indexes stopwords in documents but ignores them in queries, unless the query contains fewer than three non-stopwords.

How are simple queries processed?

Short queries consist of four or fewer non-stopwords. Documents containing all of the query words are presented in the top (fully matching) tier of results. Documents containing only some of the query words are presented in a subsequent (partially matching) tier. Within each group, documents are ranked by relevance, which takes into account the relative rarity of the query terms, the frequency with which they occur in the document and the length of the document. In addition, the ranking of documents containing the query words in their title or in the URL of the page are boosted.

Long queries consist of five or more non-stopwords. Top-tier answers need not necessarily contain occurrences of all query words. Instead, documents are ranked by relevance. If queries are very long, Funnelback will process the words in order of decreasing rarity and may ignore the most common words.

Search Fields

• Anywhere: the Anywhere field is a full-text search that also searches article metadata.
• Author: the Author field is a metadata search for article authors. Author names should be entered beginning with the last name, with quotation marks. Wildcards can be used if unsure of the full name, e.g. "Belanger, M*" or "Smith John*". The Author field is a mandatory search parameter, i.e., if an author name is entered in the search query, there should be no partial results containing articles by other authors, even if the articles match other search criteria.
• Title: the Title field searches for words contained only within the article title.
• Abstract: the Abstract field searches for words contained within the article abstract.
• Published: the Before and After fields search for articles published respectively before or after the date indicated. If both a before and an after date are specified, the search return documents published within the date range. The Before and After fields are mandatory fields, so there should be no partial results that do not match the specified date(s), even if the articles match other search criteria.
• Volume: the Volume field is a mandatory field that searches for articles published only within the specified volume of a given journal. If a volume number is entered in the search query, there should be no partial results containing articles with another volume number, even if the articles match other search criteria.
• Pages: the Pages field is a mandatory field that searches for articles beginning or ending on the specified page number. If a page number is entered in the search query, there should be no partial results containing articles with other page numbers, even if the articles match other search criteria.
• Journal: the Journal field is a mandatory field that searches for articles published in the specified journal. If a journal is specified in the search query, the results should contain only documents published within that journal, even if the articles match other search criteria. To search multiple journals simultaneously, use CTRL-click to make your selection.

Accented Characters

When searching for words containing accented characters, Funnelback will return only results containing the accented characters. If the user enters a query without accented characters, Funnelback will retrieve results matching both accented and non-accented words. For example, an author search for "Bélanger Michael" would return only documents authored by Michael Bélanger, but not Michael Belanger. However, an author search for "Belanger Michael" would find documents authored by both Michael Bélanger and Michael Belanger.

Query Language Operators

Phrase searches: searches for specific phrases or words that are unusually formed should be enclosed in quotation marks. Only results matching all of the words enclosed in quotes will be considered a match. This can reduce spurious results by requiring that the component words appear consecutively and in the order specified. Note that intervening punctuation, html tags, etc. are ignored. For example, a search for "nuclear energy" will find only articles containing the exact string nuclear energy but will not find articles containing a string such as nuclear or electrical energy.

Proximity: the near (proximity) operator consists of backquotes placed around the desired search string. The near operator requires that the component words appear, in any order, within 15 words of each other. For example, a search for `nuclear energy` will match any document containing both the words nuclear and energy occurring less 15 words apart.

Truncation: truncation consists of entering a part of a word followed immediately by a wildcard or asterisk (*) that represents 0 or more letters, up to a full word. For example, the query bio* would match all words starting with bio such as biology, biochemistry, etc. or even the word bio by itself. The truncation operator works both at the end and at the beginning of words, but not in the middle. Thus, the query *fat* matches all words containing the string fat, such as grandfather, nonfat, fatal, and fat. Truncation can also work within string (phrase) searches; the query "solar * radiation" will retrieve documents containing solar radiation as well as other phrases or strings such as solar ultraviolet radiation.

Disjunction: the disjunction operator consists of braces surrounding alternate terms. For example, a query such as [nuclear atomic] reactor will find documents containing the word reactor and either of the terms nuclear and atomic. The square brackets achieve a similar effect to the OR operator in a Boolean search.

Negation operator: the negation operator consists of an exclamation mark immediately preceding a search term. For example, documents that fully match a query such as atomic !energy will contain the term atomic but no occurrence of the word energy. Unlike the mandatory exclusion operator, partial results may contain the word energy.

Mandatory exclusion operator: the mandatory exclusion operator consists of a minus sign immediately preceding a search term. For example, documents that fully match a query such as atomic -energy will contain the word atomic but no occurrence of the word energy. Unlike the negation operator, partial results will not contain the word energy.

Mandatory inclusion operator: the mandatory inclusion operator consists of a plus sign immediately preceding a search term. For example, documents that fully match a query such as atomic +energy will include both the words atomic and energy. Partial results must contain the word energy, but they will not contain the word atomic.

Example Queries

query	explanation
a:*wkin*	Truncation operators within a metadata class. This query matches words containing the string wkin within metadata class a (author metadata).
"econom* grow*"	Truncation within a phrase. This example pattern matches phrases such as economic growth and economy growing.
t:"econom* grow*"	Truncation operators within a phrase within a metadata class (Title).
t:`G* Bush`	Truncation operators within a near operation within a metadata class (Title).

Search Results

Search results are divided into fully matching and partially matching documents, ranked by relevance. The title of the document is provided, with links to the full-text article in PDF format and article abstract in html format. The author(s), journal title, ISSN, date of publication, volume, and page numbers of the article are provided, as well as a link to journal home page.

Funnelback also provides the option of including a summary or excerpt of the article, with the search term highlighted for searches in the Anywhere field. Although the summary may give an idea of the document's contents, it is not necessarily from the most relevant part of the document.