Abstract: Document constraint descriptors are obtained from user signals. User signals are received indicating a set of attribute-value relations that can apply to documents. The user signals are used to obtain equivalent logical relations, and the logical relations are used to obtain the document constraint descriptor defining a set of one or more constraints equivalent to the logical relations. The user signals can be received through user interface circuitry of various types, including display/select user interfaces or a paper user interface in which a user marks a form that is scanned. The machine can be a portable computing device, a fixed computing device, or a multifunction device. The machine can also solve the set of constraints and can use the solution to obtain document references, such as Web URLs, indicating documents that satisfy the set of constraints. The user can then request more detail, or can obtain a printed version of a document.

Abstract: Information about documents is transferred between machines in the form of constraint descriptors. A first machine can obtain a constraint descriptor that includes information about a set of constraints that documents can satisfy. The first machine can transfer the constraint descriptor to a second machine that is capable of using the constraint descriptor to access documents that satisfy the set of constraints. Similarly, the second machine can receive a data packet from the first machine, where the data packet includes an encoding of the constraint descriptor. The second machine can decode the data packet to obtain the constraint descriptor. Each of the machines can be a portable computing device or a fixed computing device. The first machine can receive user signals through its user interface, defining an attribute-value relation, and the first machine can compile a constraint using the user signals. Or the first machine can receive user signals selecting a stored constraint descriptor.

Abstract: An information retrieval system for retrieving information from multiple information sources. The information retrieval system provides for the building of dynamic queries through the use of query channels. A query channel permits the passing of attributes of the search results between different queries. The query channel can cause the automatic passing of the attributes, or it can be user controlled (breakpoints). Further, the query results may be transformed so that they are in the context of the target query (computational transformations). For example, the results may be translated or reformatted into a form utilized by the target query. Query channels are created based on graphical representations of queries and their attributes. A simple drag and drop operation, wherein an attribute is selected and dragged to the target query, is used to create the channel. A channel may be created while a query is running.

Abstract: An information retrieval system providing for cross-lingual information retrieval. The method as implemented in the system causes elementary words to be extracted from a search expression. These elementary words are translated and stemmed into the target language(s). Searches are then made for documents containing combinations of the stemmed and translated elementary words. When documents are retrieved they are verified that they contain the same linguistic structure as the initial search expression. Documents that cannot be verified are eliminated from the set of search documents returned to the user.

Abstract: An information retrieval system for retrieving information from multiple information sources. The information retrieval system provides for the building of dynamic queries through the use of query channels. A query channel permits the passing of attributes of the search results between different queries. The query channel can cause the automatic passing of the attributes, or it can be user controlled (breakpoints). Further, the query results may be transformed so that they are in the context of the target query (computational transformations). For example, the results may be translated or reformatted into a form utilized by the target query. Query channels are created based on graphical representations of queries and their attributes. A simple drag and drop operation, wherein an attribute is selected and dragged to the target query, is used to create the channel. A channel may be created while a query is running.