Classifications

• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
  • G06Q10/00—Administration; Management
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
  • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
  • G06F16/22—Indexing; Data structures therefor; Storage structures
  • G06F16/2228—Indexing structures
  • G06F16/2246—Trees, e.g. B+trees

Definitions

• the invention relates to a method and a device for assisting the construction of an electronic document group tree.
• the number of electronic documents (mails, images, web pages, texts, ...) that a user has to manage is often of a large volume and constant growth.
• a well-known method of building a document group tree is the fully manual method. The user who has a certain number of documents to classify, creates a directory tree in which he inserts these documents as he wishes. This method has the advantage of respecting the choice of the user and also makes it easy to modify the tree manually. However, this method can become very tedious when the number of documents to be listed is important.
• Other methods provide a fully automatic classification of electronic documents. For this, the documents are defined by characteristics (for example the name of the file, its type, its size, word accounts for textual documents, colorimetric measurements for images, etc.).
• the classification methods construct groups that can be structured in a tree-like form ("Ascending Hierarchical Classification” method for example) or not ("k-Means” method for example).
• the disadvantage of such automatic partitions is that they do not always correspond to an organization desired by the user. No correction is then possible and the user is forced to accept the grouping obtained, or to restart the whole process by varying the initial parameters (number of desired groups for example) for a different result.
• Automatic learning methods that are "supervised” or “semi-supervised” make it possible to take into account criteria that the user sets a priori to implement a learning mechanism.
• supervised classification the user must give labels to some of the documents he wants to group. Two documents with the same label must join the same group and vice versa.
• a supervised learning algorithm constructs a model that allows for each unlabeled document, and based on its description, to give it an appropriate label.
• a supervised method assumes that the user knows all the possible labels of the documents to be classified and thus the groups of the final organization. The user only rarely has this knowledge a priori of the structure of the classification of his data. The initial knowledge necessary for the use of such algorithms is then a strong restriction on their use for document management.
• the object of the invention is to overcome these drawbacks by proposing a method of assisting the construction of a document group tree which allows an easily modifiable hierarchical classification of documents.
• the method according to the invention helps the user to build a document group tree so that it has a minimum of interactions and that they are the simplest possible. The user only intervenes intuitively without a thorough knowledge of computers.
• the user may not know a priori the final structure and distribution of documents in this structure.
• the method according to the invention allows the user to change these classification preferences.
• the present invention provides a method that is simple to implement and user-friendly.
• the invention proposes a method of assisting in the construction of a tree structure of groups of electronic documents, the documents being defined by predetermined characteristics.
• the method being such that it comprises for a given group of documents and a level of the given tree the following steps: a) obtaining constraints defined between at least two documents of said group and storing these constraints; b) construction of subgroups according to the constraints obtained and the characteristics of the documents of said group; c) creation of a tree level corresponding to the created subgroups; d) selecting a subgroup created and iterating steps a), b), c) and d) for the selected subgroup until said tree is obtained.
• the method allows the construction of a tree structure taking into account constraints defined for each group of documents and at each level of the tree. This construction is not therefore fixed to a definition made -AT-
• the step of building subgroups comprises the following steps:
• constraints are obtained by a step of selecting documents for which a constraint must be associated and of selecting a type of constraint using a user interface.
• the step of obtaining constraints is performed by reading predefined constraints in a memory space.
• the method following a user request to move a document from a first group belonging to a first tree level to a second group belonging to a second tree level, the method also includes the preliminary steps of:
• the method includes in addition to the preliminary steps of:
• the method makes it possible to restart the construction of the tree structure automatically, taking into account the new criteria of the modification.
• the invention also relates to a device for assisting the construction of a tree of groups of electronic documents, the documents being defined by predetermined characteristics.
• the device is such that it comprises:
• the means for obtaining constraints comprise user interface means, these user interface means comprising display means in which icons representing the constraints to be obtained are selectable by the user.
• the means for obtaining constraints comprise reading means in a memory space of saved constraints.
• the device has the same advantages as the method it implements.
• the invention also relates to a multimedia electronic equipment comprising means for storing multimedia documents, this equipment comprises means for implementing the method of assisting the construction of a tree of groups of electronic documents according to the invention.
• the invention also relates to a computer program comprising program instructions adapted to the implementation of a method of assisting the construction of an electronic document group tree according to the invention as described above, when the said program is loaded and executed in a computer system.
• the invention provides a storage medium, possibly totally or partially removable, readable by a computer, storing a set of instructions executable by said computer to implement the method of assisting the construction of a group tree electronic documents according to the invention.
• FIG. 1 represents an embodiment of a device according to the invention
• FIG. 2 represents the main constituent elements of the device according to the invention
• FIG. 3 illustrates, in flowchart form, the main steps of a method of assisting the construction of a document group tree according to the invention
• FIG. 4 illustrates in detail, in flowchart form, the step of building subgroups according to the invention
• FIG. 5 schematically illustrates the steps implemented during the construction of subgroups according to the invention
• FIG. 6a illustrates a tree level and the constraints associated with the documents belonging to a group of this tree level
• FIG. 6b illustrates two other tree levels, the groups of documents and their associated constraints after the implementation of a method according to the invention
• FIG. 7 illustrates an example of a graphical interface proposed to the user
• FIGS. 8a, 8b, 8c and 8d illustrate, in the form of a diagram of the tree structure, the steps implemented during a displacement-type modification of a document in the tree according to the invention.
• FIGS. 9a, 9b and 9c illustrate the steps implemented during a modification of the creation type of another group of documents in the tree according to the invention.
• FIG. 1 represents a block diagram of an embodiment of a device for assisting the construction of a document group tree according to the invention.
• This device is for example a microcomputer 10.
• This device has a communication bus 190 to which are connected a central unit 110, a read only memory 120, a random access memory RAM 130, a screen 140, a keyboard 150, a communication interface 180 with a communication network 115, a hard disk 170 and a data logger reader 160 on a removable medium.
• the read-only memory 120 stores, among other things, the program implementing the main steps of the method according to the invention which will be described later with reference to FIGS. 3 and 4.
• the read-only memory 120 also stores the various constraints associated with the documents that are defined by the user.
• the program according to the invention is stored in a storage means.
• This storage means is readable by a computer or a microprocessor 110.
• This storage means is integrated or not to the device and can be removable.
• the program according to the present invention is transferred into the random access memory 130 which then contains the executable code of the invention as well as the data necessary for the implementation of the invention.
• the device according to the invention has a screen capable of reproducing information representative of the construction of the tree according to the invention.
• the screen is also able to display a graphical interface to the user to determine the documents for which a constraint must be associated.
• the user Via the keyboard, or other interface means such as a computer mouse, the user selects the group of documents for which he wishes to build a tree and selects the documents for which a constraint must be associated as well. than the type of constraint.
• the hard disk 170 stores the documents that are classified according to the method of the invention.
• the hard disk may alternatively store the program implementing the method of the invention described later with reference to FIGS. 3 and 4.
• the data logger reader 160 on a removable memory means is for example a recorder player for compact discs or DVDs.
• Documents classified according to the invention can also be saved on a removable storage means such as a compact disc or a DVD.
• the recording reader is able to read the program according to the invention for the transfer thereof to the hard disk 170.
• the classification of documents and the tree of groups of documents obtained by the implementation of the invention are not necessarily coupled with the physical organization of documents in the storage system.
• the representation of the tree resulting from the implementation of the invention may be only a logical representation. Thus several logical representations can be obtained for a single physical organization.
• FIG. 2 shows in block diagram form the main components of the device of the invention.
• the documents from a storage base 280 are processed by the vectorization module M210.
• This vectorization module makes it possible to construct for each document a digital vector describing this document. For example, for an image, this may consist of a set of measurements, functions applied to the bitmap representation of the image, for a text document, this may consist of a calculation of predefined words for each document. Thus, a vector will represent the characteristics of the corresponding document. It is then possible to define a distance between two vectors and thus a measure of proximity between documents. The distance between two vectors will represent the distance between the characteristics of the documents.
• the normalization module M220 makes it possible to standardize the vectors representing the documents so as to obtain vectors of constant norm. This normalization can for example be done by scaling all the descriptive variables in the interval [0; l].
• the documents thus processed by the M210 and M220 modules then arrive at the input of the device for assisting in the construction of a document tree according to the invention 200.
• This device includes, inter alia, an M240 user interface module that allows a user to work in interactive mode on document classifications.
• This module notably allows the user to define constraints to be associated with the documents, to modify the tree built according to the method of the invention.
• the device also comprises an M250 module for obtaining constraints defined between at least two documents of a group. Obtaining constraints can be done via the GUI module M240 or for example by simply reading constraints previously defined and stored in a memory space of the device (M280). The newly obtained constraints are stored in a memory space M280 of the device.
• the device comprises a module M260 for building document subgroups according to the constraints obtained by the module M 250 and characteristics specific to the documents. It includes a module for creating M290 tree levels corresponding to the subgroups from the M260 module. The M290 module is linked to the user interface module for displaying the resulting tree.
• the device comprises an M270 module for selecting a group or subgroup created by the module M260.
• the selection of a group or subgroup is made via the user interface module.
• the group thus selected will then be applied to the previously described modules of the device in order to obtain at the output of the device 200, a document group tree that satisfies the user.
• the device according to the invention can be integrated in a multimedia electronic equipment which comprises means for storing multimedia documents such as images, music files, written documents, etc.
• This equipment is for example a music file reader.
• This equipment also includes a graphical interface, a pocket computer or electronic organizer, a mobile phone with or without a camera.
• the method for assisting the construction of a tree structure according to the invention can be implemented on this type of equipment for classifying the stored multimedia contents.
• a step E300 obtaining constraints between at least two documents of the group is performed.
• This obtaining step is for example the selection by the user of two documents and a graphic interface icon representing a constraint of the type "must be associated” or "Must-Link” in English (ML) or of type “must not be associated” or “Can not-Link” in English (CL).
• ML Must be associated
• ML Must-Link
• CL Can not-Link
• the constraint of type ML expresses the will to find the two documents in the same group at the lower level (subgroup) and the constraint of type CL expresses the will not to find the two documents in the same group at the lower level.
• a constraint is defined by a pair of documents, a type of constraint (ML or CL) and a level of the tree. These constraints thus defined are stored in a memory space of the device.
• Obtaining constraints in step E300 can also be performed by simply reading constraints previously stored in a memory space.
• constraints can also be carried out by the selection by the user of a certain number of documents that he wishes to associate in the same subgroup. He can thus create a draft of subgroups. Constraints are then automatically created in step E300, the set of documents representative of the subgroups being associated two by two (according to all the possible pairs of two different documents) by constraints CL.
• step E300 The constraints thus defined in step E300 are then used in step E310 for the construction of subgroups.
• the construction of these subgroups is also a function of the characteristics of the documents.
• the subgroup construction step will be detailed later with reference to FIGS. 4 and 5.
• Step E310 is followed by step E320 where the creation of a new tree level corresponding to the subgroups thus created is performed.
• a representation of the tree being created is shown in Figures 6a and 6b.
• step E330 the user can stop the creation of the tree if he considers that it is satisfactory. In this case (O), the process is finished. In the opposite case (N), the step E330 is followed by the step E340 where the selection of a subgroup previously created and at the lower level of the tree is performed. This selection can be made by the user via the graphical interface.
• a criterion which can be for example the selection of the subgroup of the largest population or the subgroup containing the last selected image.
• Step E340 is followed by step E300 previously described.
• the steps E300, E310, E320, E330 and E340 are thus iterated for the selected subgroup until a satisfactory tree for the user is obtained.
• FIG. 4 illustrates in detail the step E310 described with reference to FIG. 3.
• step E400 consists in creating a group per document participating in at least one constraint.
• step E400 is illustrated in FIG. 5 where the documents are represented by solid circles, the constraints ML by solid lines and the constraints CL by dashed lines. At this stage, a group has only one document (a kernel). Step E400 is followed by step E410 where an association of the documents not participating in at least one constraint with the groups created in step E400 is performed.
• This association is carried out according to a criterion of distance between the characteristics of the documents to be associated and those of the documents belonging to the groups.
• the distance of a document to a newly created group is defined as the distance between the document and the group's kernel (at this point, each group has a unique kernel).
• the criterion of associating a document with a group is a criterion of the smallest distance.
• the distance between document characteristics is achieved by a distance measurement between vectors representing the characteristics of the documents.
• step E410 groups are obtained as illustrated under the reference of step E410 in FIG. 5.
• Step E410 is followed by step E420 where an iterative merger of certain groups obtained in the previous step is performed.
• This merge is performed according to the constraints associated with the documents.
• the two groups merge automatically. This is illustrated as E420 in Figure 5.
• Step E420 is followed by step E430.
• step E430 a group merge is performed so that if a group G2 and a group G3 as illustrated in FIG. 5 do not respectively comprise a document for which there exists a constraint CL, then G2 and G3 merge.
• the subgroups illustrated under the reference E430 of FIG. 5 are then obtained.
• each subgroup has one or more nuclei.
• Step E430 is followed by step E320 previously described with reference to FIG. Figures 6a and 6b illustrate the construction of the document groups of the tree with the constraints associated with each level of the tree.
• FIG. 6a represents for example a group of documents that corresponds to the documents illustrated in FIG. 5 under the reference E400. Constraints on some of these documents have been defined and recorded and are shown in Figure 6a. These constraints remain associated with the corresponding level, here the level 0.
• Figure 6b illustrates the creation of two tree levels.
• the first level created (level -1) corresponds to the subgroups obtained at the end of the step E430 illustrated with reference to FIG. 5.
• the creation of this tree level corresponds to the step E320 described with reference to FIG. Figure 3.
• step E340 of FIG. 3 The selection of a group of documents corresponding to step E340 of FIG. 3 is shown in negative in FIG. 6b.
• the steps E300, E310, E320 are executed again.
• New constraints corresponding to the level -1 are thus defined and represented in FIG. 6b, new subgroups are created and another level of tree structure (level -2) is created and represented in FIG. 6b.
• Figure 7 is an illustration of an example of a graphical user interface presented to the user.
• a document group in the tree represented in window 700 is selected. This selected group is shown in negative.
• the documents of this selected group are displayed in a second window 710.
• the user can then select two documents, for example documents dl7 and dl9 and then select the icon representing the constraint he wishes to associate with these documents, for example the icon 770 corresponding to the ML constraint.
• a constraint associated with the level -2 of the tree is then created and displayed in the third viewing window 720.
• the constraint CL can be selected by the corresponding icon 770.
• the The user may select the "learn" icon 730 so that step E310 of the method shown in FIG. 3 is executed.
• a tree structure When a tree structure has been partially constructed according to the method of the invention, it can be modified at any time using the same method.
• the user wants to delete a part of the tree, he selects for example a document group of the tree using for example the graphical interface of Figure 7. He then "clicks" on the icon 760 to delete the entire lower part of the tree. By this action, a step of moving in the selected group of group documents of the part of the deleted tree is performed. All document constraints for lower levels of the tree are also removed.
• FIG. 8a the user wishes to move the document d from a level group -2 in the tree to a level group -3.
• the user for example moves with the mouse of the icon representing the document to be moved and selected in the window 710 of the graphical interface of FIG. 7, from the first group of documents to the second group.
• the tree level constraints are saved in memory.
• the method that will allow to rebuild a different tree will begin by first determining and selecting the lowest common group top level in the tree, ie the first group common to the initial group of d and destination group of the move, encountered when going up the tree from the two groups concerned, to the higher levels. Any part of the level tree below this selected group is removed. As mentioned earlier for deleting a part of the tree, all the documents from the deleted part's groups are moved to the selected group. However, in this case, the constraints of the lower levels which concern other documents than the one to be moved are kept in memory. New constraints are defined to take into account the displacement of the document d.
• the constraints represented underlined have been created automatically. They express the fact that at level -1, the document d and the core document dl must be grouped together (ML "Must link") in order to force d to join the subgroup of dl at level -2 (FIG. 8c). . The same goes for document d and the d5 kernel document at level -2.
• the tree construction method described with reference to FIG. 3 is implemented from the selected upper common group and with the constraints already defined and saved in memory. After the application of this method, a tree is obtained as shown in Figure 8d where the document has been moved to the desired group. Some documents close to the moved document that do not participate in constraints and that were in the same group as d will automatically follow the same path as document d and will end up in the same destination group.
• New CL type constraints are created as shown in underlined with reference to Figure 9b. These new constraints specify that the document d must be separated from the kernel documents dl and d4 at level -2. As d can not anymore join any of the subgroups, a new subgroup will be automatically created during a construction phase of the tree defined with reference to Figure 3. Note that, in order to prevent d from being able to join the sub -group containing the kernels dl and d5 at -2, the new constraint CL (dl, d) was created at level -1. The constraint CL (d5, d) would equally play this role.
• the user can maintain his file management habits while being assisted by the implementation of an automatic method of building a tree structure, via the automatic or manual definition of ML and CL links between a small number of files. documents. This brings the user a significant time saving and satisfaction with the result obtained.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Business, Economics & Management (AREA)
• General Physics & Mathematics (AREA)
• Physics & Mathematics (AREA)
• Marketing (AREA)
• Entrepreneurship & Innovation (AREA)
• Quality & Reliability (AREA)
• Strategic Management (AREA)
• Tourism & Hospitality (AREA)
• Human Resources & Organizations (AREA)
• General Business, Economics & Management (AREA)
• Operations Research (AREA)
• Economics (AREA)
• Software Systems (AREA)
• Data Mining & Analysis (AREA)
• Databases & Information Systems (AREA)
• General Engineering & Computer Science (AREA)
• Document Processing Apparatus (AREA)
• Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Applications Claiming Priority (2)

Publications (1)

Family

ID=37308901

Family Applications (1)

Country Status (4)

Families Citing this family (8)

Family Cites Families (6)

• 2006
  • 2006-01-03 FR FR0650024A patent/FR2895813A1/fr active Pending
  • 2006-12-20 WO PCT/FR2006/051402 patent/WO2007077378A1/fr active Application Filing
  • 2006-12-20 EP EP06847192A patent/EP1984873A1/de not_active Ceased
  • 2006-12-20 US US12/087,506 patent/US8886678B2/en active Active

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events