Glossary

• AIAT  

  See Apple Information Access Toolkit (AIAT).



• Apple Information Access Toolkit (AIAT)  

  In Classic Mac OS, an object-oriented information access engine that contained a collection of tools for indexing, searching, and analyzing large volumes of documents. Search Kit is the OS X implementation of the AIAT. AIAT was formerly known by its code name V-Twin.



• Boolean searching  

  Matching of a query string to indexed terms using Boolean (logical) operators such as AND and OR between query terms, optionally employing grouping for precedence using parentheses. The entire query expression is matched. See also search.



• compact  

  To make an index smaller by removing unused bits. Over time, as documents get added to and removed from an index, the index’s disk or memory footprint may grow due to fragmentation. Search Kit includes APIs to check for fragmentation and to compact an index. See also fragmentation .



• corpora  

  Plural form of corpus.



• corpus  

  A collection of one or more documents, typically related, and available to an information retrieval system. Plural: corpora.



• document  

  In general, a specifically locatable information object of useful granularity and arbitrary structure. In Search Kit, anything that contains text and that the Search Kit client application addresses as a document—an RTF document, a PDF file, a Mail message, an Address Book entry, the contents at an Internet URL, the result of a database query, and so on. See also document URL object.



• document collection  

  See corpus.



• document object hierarchy  

  A collection of documents in which each document exists at a location relative to a root document. The locations may may be real, as in a file system, or virtual.



• document URL object  

  A URL to a document. In Search Kit, a document URL object comprises a scheme, a parent document URL object, and a name, with the format of each component defined by the client application. Search Kit document URL objects may be converted to or from CFURL objects. See also document, parent document URL object, scheme.



• fragmentation   

  In Search Kit, an unwanted increase in index size due to accumulation of unused capacity. Over time, as documents get added to and removed from an index, the index may become fragmented—its constituent documents and terms may become arranged in a manner that includes a significant amount of unused disk or memory space. See also compact.



• inclusion/exclusion result  

  See inclusion/exclusion searching.



• inclusion/exclusion searching  

  Unranked searching where the result simply includes documents that match the query and excludes documents that don’t. Inclusion/exclusion searches tend to be faster than ranked searches. Search Kit supports inclusion/exclusion searches. See also relevance-based result.



• index  

  A memory- or file-based sequential collection of the terms in one or more documents. In addition to terms, Search Kit indexes contain context information that specifies which documents each term belongs to, along with term and document metadata useful during display of search results. Search Kit performs its searching and analysis on indexes. See also inverted index; inverted-vector index; vector index



• index group  

  A short-lived collection of one or more indexes; the target of a search. An index group corresponds to one or more aspects of the corpus of documents you want to search. For example, one index in a group might contain document titles, while another contains the body text of those same documents. An index group can also comprise indexes of multiple corpora. See also corpus; document.



• information retrieval (IR)  

  The process of locating information based on a well-defined information need. An information retrieval system consists of a corpus, one or more indexes of its content, a query interface, a search system, and a results interface. See also corpus; search.



• inverted index  

  An index containing terms, as keys, mapped to references to the documents they appear in. The index is sorted by its keys. “Inverted” means that the documents are found by matching on terms, rather than the other way around. See also index; inverted-vector index; vector index



• inverted-vector index  

  An index containing terms mapped to document URL objects representing the documents that the terms appear in, as well as document URL objects mapped to the terms that each document contains. See also index; inverted index; vector index.



• IR  

  See information retrieval (IR).



• MIME type hint  

  Advisory metainformation suggesting the likely content type for a URL. MIME is an acronym for Multipurpose Internet Mail Extensions. In Search Kit, common MIME type hints include text/plain, text/rtf, text/html, text/pdf, and application/msword.



• minimum term frequency  

  The fewest number of times a term can appear in a document and still be indexed. This functionality is not currently supported by Search Kit indexes.



• minimum term length  

  The shortest-length term to index. When Search Kit adds terms from a document to an index, it skips over words whose length is shorter than the minimum term length.



• name  

  In Search Kit, a document name as represented in a document URL object. For documents that are on-disk files, the name should correspond to the actual filename. For other types of documents, your application can assign any name to a document. See also document URL object



• operator  

  A character or word that has a special meaning when used in a query. Operators in Search Kit include AND, OR, NOT, parentheses, quotation marks, and several others. Search Kit interprets operators and determines the user's intended search type according to the operators' meanings.



• parent document URL object  

  In Search Kit, for file-based documents, the location of the enclosing folder for a document or for another parent document URL object. Search Kit manages documents using parent-child relationships, not paths. You can construct the path of any document by following its parent document links. See also document URL object.



• partial string searching  

  Matching of the terms in a query string to indexed terms, with implied wildcard characters at the start and end of each query term. Each term is matched separately. Search Kit does not currently support partial string searching as an option, but a client application can provide it by adding wildcard operators (asterisks) around each term before handing a query off to Search Kit. See also search.



• phrase searching  

  Matching of a query string to indexed terms, with the query string considered as a complete phrase. A match occurs when the exact query phrase appears in a document. Search Kit supports phrase searching in inverted and inverted-vector indexes. See also search.



• prefix searching  

  A specialized type of substring search. A prefix search involves matching of a term in a query string to indexed terms, with an explicit wildcard character at the end of the query term. A match occurs when the characters in the query term (minus the wildcard character) match the beginning of an indexed term. For example, the query string car* will match car, carpet, and carnivore. Search Kit supports prefix searching in inverted and inverted-vector indexes. See also search; substring searching; wildcard character.



• query  

  (n.) A text string, containing terms and operators, that represents a user's information retrieval request. Various types of query supported by Search Kit include simple, prefix/suffix/substring, Boolean, phrase, and similarity. (v.) To invoke a request for information in an information retrieval system. See also search.



• ranked searching  

  See relevance-based result.



• relevance-based result  

  See also relevance-based search.



• relevance-based search  

  A ranked search whose result includes a relevance rating for each document matching a query. In general, relevance ratings may be normalized to 100%, or nonnormalized. Search Kit supports only nonnormalized results. See also inclusion/exclusion searching; search.



• root word  

  See stem.



• scheme  

  A way to access a file-system or Internet resource, corresponding to an access protocol. Examples include http, ftp, and file. See also document URL object.



• search  

  In an information retrieval system, a process that attempts to locate documents that match a query, and that may assign relevance scores to the found documents. Upon a successful match, a search system returns references to the found documents. Search Kit supports a variety of search types, some of which can be combined. These types are simple, Boolean, ranked, unranked, phrase, similarity, prefix, suffix, and substring.



• search object  

  In Search Kit, an opaque data type representing an asynchronous search and containing its results, accumulated as they are found. A search object is of type SKSearchRef.



• similarity searching  

  Matching of a query string, typically consisting of a representative portion of a document, to indexed documents. A match occurs when Search Kit determines significant content similarity between the query and an indexed document. Search Kit supports similarity searching in vector and inverted-vector indexes. Similarity searching also works in inverted indexes in Search Kit, but performance is worse. See also search.



• simple search  

  Matching of the terms in a query string to indexed terms using exact, character-for-character matching. Each term is matched separately. In Search Kit, by default, spaces between terms behave like Boolean AND operators. See also search.



• stem  

  The root of a family of morphological or inflectional variants of a word. For example, "swim" is the stem of "swimmer," "swimming," and "swam."



• stemming  

  The algorithm-based removal of morphological and inflectional word components, typically endings. Language dependent. Stemming is sometimes referred to as suffix stripping, although some stemming algorithms perform prefix stripping as well. IR systems use stemming to improve search quality and to reduce index size. Search Kit does not support stemming; if needed, client applications implement it. Some stemming algorithms handle only regular variants, such as converting "swimming" to "swim," and do not handle irregular variants, such as converting "swam" to "swim."



• stopword  

  A word not to index. When Search Kit adds terms from a document to an index, it skips over words in its top-word list.



• substring searching  

  Matching of a term in a query string to indexed terms, with explicit wildcard characters at the start and end of the query term. A match occurs when the characters in the query term (minus the wildcard characters) match the beginning, ending, or middle of an indexed term. For example, the query string *cat* will match cat, concatenate, tomcat, and cattle. Search Kit supports substring searching in inverted and inverted-vector indexes. See also search.



• suffix searching  

  A specialized type of substring search. A suffix search involves matching of a term in a query string to indexed terms, with an explicit wildcard character at the start of the query term. A match occurs when the characters in the query term (minus the wildcard character) match the ending of an indexed term. For example, the query string *ion will match ion, lion, and version. Search Kit supports suffix searching in inverted and inverted-vector indexes. See also search; wildcard character.



• suffix stripping  

  See stemming.



• summarization object  

  In Search Kit, an opaque data type representing summarization information, including the summary text. A summarization object is of type SKSummaryRef.



• synonym  

  A term that an IR system considers to be equivalent to another term for both indexing and querying. For example, an IR system could define "car," "passenger vehicle," and "automobile" to be synonyms. See also information retrieval (IR); index; query.



• term  

  An atomic entry in a Search Kit index, typically corresponding to a word found in one of the index’s documents.



• text extraction  

  Selective copying of terms from one or more documents into an index. See also stemming; stopword.



• unranked searching  

  See inclusion/exclusion searching.



• URL  

  Uniform Resource Locator. An Internet address, or a file-system path when formatted as a URL with a scheme. See also scheme.



• V-Twin  

  See Apple Information Access Toolkit (AIAT).



• vector index  

  An index containing document URL objects, as keys, mapped to the terms that each document contains. See also index; inverted index; inverted-vector index



• wildcard character  

  An operator used in a query that indicates matching on any character. In Search Kit, the wildcard character is the asterisk. Depending on usage, the wildcard character can indicate prefix, suffix, or substring searching. See also operator; query.