JP2007304692A - Event retrieval device, method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To retrieve an event and a phenomenon caused in the real world by a word and a sentence expressed by natural language. <P>SOLUTION: Measurement data from respective sensors installed on an object and time information on the measurement data, are stored as a set by a sensor DB 11. The corresponding relationship between the word indicating the event on the object, an event descriptor indicating a group belonging with the word and physical quantity expression of the event expressed by an expression using the measuring data, is stored by an event descriptor dictionary 13. The event descriptor corresponding to the predicate is retrieved from the event descriptor dictionary 13 by extracting the predicate indicating the event being an retrieval object from question information input by the natural language by a question accepting part 12. The measurement data corresponding to the physical quantity expression corresponding to its event descriptor, is retrieved from the sensor DB 11 by a data retrieval part 10, and retrieval result information on the event being the retrieval object is output. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a sensor network technology, and more particularly to an event search technology for detecting an event or phenomenon corresponding to a word or sentence expressed in a natural language.

  Environmental conditions, both indoors and outdoors, change from moment to moment depending on the people present or other factors. Some of them can be detected by sensors, and by gathering information from a sensor network in which multiple sensors with network connection functions are distributed in the environment, it is possible to grasp changes in the environment as a whole or in part. It becomes. The sensor here refers to a device that detects a physical quantity or signal in the vicinity.

  In particular, events and phenomena caused by people living in the environment and those existing there cause changes in the environment. If this situation change is recorded and accumulated by a sensor installed in the environment, it becomes possible to identify events and phenomena that have occurred in the past. Furthermore, by expressing events and phenomena with natural language words and syntax, it is possible to construct an event / phenomenon search apparatus with high affinity for humans.

  According to such an apparatus, it is possible to search for an event and predict a situation having a high probability of occurring next. Also, as a part of the situation, if it is possible to predict the action performed by the user who is in the situation, the user's danger situation will be detected and a warning will be issued to avoid or the action performed by the user will be performed. The system can be implemented instead. In fact, automatic doors and HVAC systems can be thought of as a simple alternative to the latter, using a combination of sensors and actuators.

  In addition, many research results on home appliance networks and sensor networks have been announced in connection with ubiquitous computing, and application applications to crime prevention, medical sites, and in-vehicle devices have already begun (for example, see Non-Patent Document 1). These achievements are largely due to improved computer performance, the emergence of new network standards, and the miniaturization of batteries, power supplies, and sensor devices. As a result, conventional methods for recognizing situation changes that occur in the environment are only methods that rely heavily on image recognition using cameras fixed to walls, ceilings, etc., and on installation conditions that rely on sensors with fixed observation targets. Rather, the conditions for constructing a device that installs a large number of sensors in the environment and searches for events and phenomena occurring in the environment based on the information collected there have been established.

For example, in the SmartDust project shown in Non-Patent Document 2, an idea has been proposed in which an ultra-small sensor node is mixed with paint and applied to a wall to recognize the surrounding situation. In the Smart-ITS project, a medicine management system that issues a warning not to place a medicine that causes a severe reaction next to the sensor node is constructed based on the proximity relationship between sensor nodes.
In addition, it is a well-known fact that web searches using natural language words are performed on a daily basis. In addition, as shown in Non-Patent Document 3, a system has been developed that asks where a sentence close to a natural language is present in the environment.

Y. Nakauchi, K. Noguchi, P. Somwong, T. Matsubara, and A. Namatame. Vivid room: human intention detection and activity support environment for ubiquitous autonomy, Proc. Of the 2003 IEEE / RSJ International Conference on Intelligent Robots and Systems (IROS2003), PP.773-773, 2003. J.M.Kahn, R.H.Katz, and K.S.J.Pister., Next century challenges: mobile networking for smart dust, Proc.MOBICM, pp.271-278, 1999. K. Yap, V. Srinivasan, and M. Motani.MAX: human-centric search of the physical world, Proc. Of the 3rd ACM Conference on Embedded Networked Sensor Systems (SenSys'05), pp.166-179 , 2005. C. Fellbaut, WordNet: an electronic lexical database, MIT Press, 1998. Satoshi Hiramatsu, Masami Hattori, Atsumi Yamada, Tsuyoshi Okadome. Pervasive Association: Semantic Integration in Ubiquitous Environment (2) Collecting Real-World Relational Information Using One Sensor Network 1 Multimedia, Distributed, Cooperation and Mobile (DlCOMO2005) Symposium Paper Shu, pp.169-172, 2005. Sadaaki Furui, Acoustics and Speech Engineering, Modern Science, 1992.

However, in such a conventional technique, in an environment where a large number of sensors that sequentially record changes in the situation are installed, it is possible to detect the environment observed by the sensor and the state change of the object in the environment, but the words expressed in natural language As in the case of searching information on the Web by using text or sentence, it is not possible to search for events and phenomena that occur widely in the real world.
The present invention is intended to solve such problems, and provides an event search apparatus, method, and program capable of searching for an event or phenomenon that has occurred in the real world using words or sentences expressed in natural language. It is an object.

  In order to achieve such an object, the event search apparatus according to the present invention searches the sensor database storing the measurement data measured by a plurality of sensors attached to the object in the environment, and thereby in the environment. An event search device that searches for any event that has occurred, and includes measurement data measured by a plurality of sensors attached to an object in the environment, and time information indicating the time at which the measurement data was measured. Correspondence between a sensor database stored as a set, a predicate consisting of a verb indicating an event related to an object, an event descriptor indicating a group to which the predicate belongs, and a physical quantity expression of the event expressed by an expression using measurement data An event descriptor dictionary that memorizes the relationship and a predicate indicating the event to be searched are extracted from the question information input in natural language, and the predicate corresponding to the predicate is extracted from the event descriptor dictionary. Search the event database for the question reception unit that searches for event descriptors, and search the sensor database for measurement data corresponding to the physical quantity expression corresponding to the event descriptor obtained by the question reception unit. A data search unit that outputs search result information.

  At this time, in the question reception unit, the object information indicating the object related to the event to be searched is extracted from the question information, and when the measurement data is searched in the data search unit, the measurement data corresponding to the physical quantity expression is used. In addition, measurement data obtained from an object that matches the object information may be retrieved from the sensor database.

  In addition, predicate modifiers consisting of adverbs, prepositions or adjectives indicating temporal or spatial conditions related to events, auxiliary descriptors indicating the groups to which these predicate modifiers belong, and expressions using measurement data An auxiliary descriptor dictionary that stores the correspondence between physical quantity expressions of conditions and predicate modification consisting of adverbs, prepositions, or adjectives that indicate temporal or spatial conditions related to the event from the question information in the question reception unit The word is extracted, the auxiliary descriptor corresponding to the predicate modifier is searched from the auxiliary descriptor dictionary, and the combined descriptor and auxiliary descriptor obtained by the question reception unit when the measurement data is searched by the data search unit Measurement data corresponding to each physical quantity expression corresponding to may be retrieved from the sensor database.

  In addition, the physical quantity expression includes a time constraint expression part including a time constraint expression indicating each section before, during and after the occurrence of the event, and a physical expression including a physical expression expressing the event in the section with an arbitrary physical quantity. You may comprise from the group with a formula part.

  Further, a syntax analysis unit that analyzes the syntax of the question information may be further provided, and the predicate may be extracted by the question reception unit based on the analysis result of the syntax analysis unit.

  In addition, for each event descriptor, an event reference database that stores a data pattern indicating a change in measurement data related to the object is further provided, and the event description obtained by the question reception unit when the measurement data is searched by the data search unit A data pattern corresponding to the child may be searched from the event reference database, and measurement data that changes in a similar manner to the data pattern may be searched from the sensor database.

  Each event descriptor further includes an event reference database that stores segment information for specifying measurement data related to the event, and the data search unit obtains the event descriptor obtained by the question reception unit when searching the measurement data. May be retrieved from the event reference database, and measurement data corresponding to the physical quantity expression corresponding to the event descriptor may be retrieved from the measurement data corresponding to the segment information.

  In addition, the event search method according to the present invention searches for a sensor database that stores measurement data measured by a plurality of sensors attached to an object in the environment, thereby searching for any event that has occurred in the environment. An event search method used in an event search apparatus, wherein measurement data measured by a plurality of sensors attached to an object in an environment by a sensor database, and time information indicating a time when the measurement data was measured , And the event descriptor dictionary, a predicate consisting of a verb indicating an event related to the object, an event descriptor indicating the group to which the predicate belongs, and the event represented by an expression using measurement data Storing a correspondence relationship between the physical quantity expression and the question reception unit, and a question reception unit that determines a phenomenon to be searched from question information input in a natural language. The predicate is extracted, the event descriptor corresponding to the predicate is searched from the event descriptor dictionary, and the measurement data corresponding to the physical quantity expression corresponding to the event descriptor obtained by the question reception unit is obtained by the data search unit. A step of searching from the sensor database and outputting search result information relating to an event to be searched according to the search result.

  In addition, the program according to the present invention searches for a sensor database that stores measurement data measured by a plurality of sensors attached to an object in the environment, thereby searching for any event that has occurred in the environment. Storing, as a set, measurement data measured by a plurality of sensors attached to an object in the environment and time information indicating a time at which the measurement data was measured by a sensor database in the computer of the apparatus; Correspondence between a predicate consisting of a verb indicating an event related to an object, an event descriptor indicating a group to which the predicate belongs, and a physical quantity expression of the event expressed by an expression using measurement data And a predicate indicating an event to be searched from the question information input in natural language by the question receiving unit, A step of retrieving an event descriptor corresponding to the predicate from the elephant descriptor dictionary, and a data retrieval unit to retrieve measurement data corresponding to the physical quantity expression corresponding to the event descriptor obtained by the question reception unit from the sensor database; A step of outputting search result information relating to an event to be searched according to the search result.

  According to the present invention, the sensor database stores measurement data measured by a plurality of sensors attached to an object in the environment and time information indicating the time when the measurement data is measured as a set. At the same time, the event descriptor dictionary stores a correspondence relationship between a word indicating an event related to an object, an event descriptor indicating a group to which the word belongs, and a physical quantity expression of the event expressed by an expression using measurement data. Then, the question receiving unit extracts a predicate indicating an event to be searched from the question information input in the natural language, and searches the event descriptor corresponding to the predicate from the event descriptor dictionary.

Then, the data retrieval unit retrieves the measurement data corresponding to the physical quantity expression corresponding to the event descriptor obtained by the question reception unit from the sensor DB, and the retrieval result information regarding the event to be retrieved is obtained according to the retrieval result. Is output.
This makes it possible to search for events and phenomena that have occurred in the real world using words and sentences expressed in natural language in an environment in which a large number of sensors that sequentially record changes in the situation are installed.

Next, embodiments of the present invention will be described with reference to the drawings.
[First Embodiment]
First, an event search apparatus according to a first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing the configuration of the event search apparatus according to the first embodiment of the present invention.
The event search device 1 is composed of an information processing device such as a computer that performs arithmetic processing on input information, and searches a sensor database that stores measurement data measured by a plurality of sensors attached to objects in the environment. To search for any event that occurred in the environment.

  In the present embodiment, a sensor database that stores measurement data measured by each sensor and time information indicating the time at which the measurement data was measured as a set, a word indicating an event related to an object, and the word The event descriptor dictionary that stores the correspondence between the event descriptor indicating the group to which the group belongs and the physical quantity representation of the event expressed by an expression using measurement data is provided, and the search target is based on the query information entered in natural language Search for event data indicating the group to which the word belongs from the event descriptor dictionary, and search the sensor database for measurement data corresponding to the physical quantity expression corresponding to the event descriptor. According to the search result, search result information related to the event to be searched is output.

Next, the configuration of the event search device according to the first exemplary embodiment of the present invention will be described in detail with reference to FIG.
The event search apparatus 1 includes a data collection unit 10, a sensor database (hereinafter referred to as sensor DB) 11, a question reception unit 12, an event descriptor dictionary 13, a data search unit 14, and a question response unit 15 as main functional units. Is provided.

  Among these functional units, the data collection unit 10, the question reception unit 12, the data search unit 14, and the question response unit 15 may be realized by a dedicated arithmetic processing circuit unit, but a microprocessor such as a CPU and its peripheral circuits The functional unit may be realized by cooperating the hardware and the program by providing an arithmetic processing unit consisting of the above, reading a program prepared in advance, and executing it by a microprocessor.

  The sensor DB 11 and the event descriptor dictionary 13 are realized by a storage unit (not shown) including a storage device such as a memory and a hard disk, and processing information exchanged between the functional units is also exchanged via the storage unit. Is done. The program used in the arithmetic processing unit may be stored in the storage unit. In addition, the event search device 1 is provided with various basic configurations such as an operation input device, a screen display device, and a data input / output device, as in a general information processing device. Used.

The data collection unit 10 has a function of collecting measurement data and attribute information measured by each sensor node 40 (41 to 44) arranged in the environment 3 via the sensor network 2, and the collected measurement data and attribute information. Is stored in the sensor DB 11.
The sensor node 40 has a sensor that detects and outputs an event occurring in the environment 3, such as an acceleration sensor, a temperature sensor, and a human sensor, and a function of performing data communication wirelessly or by wire.

These sensor nodes 40 are attached to objects such as a table 31, a chair 32, a vase 33, and a book 34 disposed in the environment 3 such as a room, and in addition to the measurement data measured by the sensor, the measurement data Attribute information such as time information indicating the time at which the sensor node 40 is measured, a sensor ID for identifying the sensor used for the measurement, and the name of the object to which the sensor node 40 is attached are output.
FIG. 2 is an explanatory diagram illustrating a configuration example of measurement data. Here, accelerations relating to the table 31, the chair 32, the vase 33, and the book 34, which are objects, are measured by sensors and output as measurement data.

The sensor DB 11 is a database that stores a set of measurement data and attribute information thereof.
FIG. 3 is an explanatory diagram illustrating a configuration example of the sensor database. Here, “sensor ID” of the sensor from which the measurement data is detected, “object information” indicating the name and identification information of the object to which the sensor node is attached, “data type” indicating the content of the measurement data, “Data unit”, “measurement data”, and “time information” of the measurement data are stored as a set.

  The question receiving unit 12 includes, as words from the question information input in a natural language, a predicate including a verb indicating an event to be searched, an adverb indicating a temporal or spatial condition related to the event, a preposition, or an adjective. It has a function of extracting a predicate modifier and a function of retrieving an event descriptor and a physical quantity expression corresponding to the predicate from the event descriptor dictionary 13.

The event descriptor dictionary 13 includes a predicate consisting of a verb indicating an event related to an object, an event descriptor consisting of a label indicating a group to which the predicate belongs, a physical quantity expression of the event expressed by an expression using measurement data, It is a database that stores the correspondence relationship.
FIG. 4 is an explanatory diagram showing a configuration example of the event descriptor dictionary. Here, “predicate”, “event descriptor”, and “physical quantity expression” indicating an event related to the object are stored in association with each other.

The data search unit 14 has a function of searching the sensor DB 11 for measurement data corresponding to the event descriptor and the physical quantity expression based on the event descriptor and the physical quantity expression searched from the event descriptor dictionary 13 by the question receiving unit 12. In addition, it has a function of outputting the measurement data and its attribute information as search result information regarding an event to be searched according to the search result.
The question answering unit 15 receives search result information from the data search unit 14 and outputs it as an answer corresponding to the question information received by the question receiving unit 12 from a screen display device (not shown) or a data input / output device. have.

[Event descriptor dictionary]
Next, the event descriptor dictionary will be described in detail with reference to FIG. FIG. 5 is an explanatory diagram showing a specific example of the event descriptor dictionary.

The event descriptor dictionary 13 is generated in advance as follows. First, using the thesaurus system, the basic verbs related to “movement” and the basic verbs related to “scalar physical quantity” are used as seeds to follow the links related to the thesaurus, and to deal with physical phenomena. Collect as many predicates as possible.
Here, using the English electronic thesaurus system WordNet (see Non-Patent Document 4 etc.), the verb “move, reach, pass, exit, touch, enter” related to “movement” and “scalar physical quantity” Using the verbs “rise, drop, increase, keep, remain” as seeds, we follow links in the thesaurus one after another and collect all predicates that can be associated with physical phenomena.

  At this time, a plurality of predicates having exactly the same meaning are labeled as one group by using a clause expressing the meaning described in WordNet. For example, the verb “shift, dislodge, reposition” forms one group with a common meaning “change direction” and labels the group “change-direction”. In this way, each labeled predicate group is created.

  Next, for each of these predicate groups, a distinction is made between transitive verbs and intransitive verbs, and auxiliary symbols are introduced into the labels. For example, for the label “change-location”, in the case of the group of the transitive verb “move, travel, locomote, go”, two arguments are added like “change-location (a, o)”, and the argument “a” in the transitive verb Represents the subject and allows the argument o to represent the object. On the other hand, in the case of an intransitive verb, only “change-location (o)” and the subject can be clearly indicated, or it is omitted and expressed as “change-location”.

Furthermore, there is a word that includes information such as time and place as its meaning, and an expression that adds an argument that specifies the information to the label for the word group is introduced. In other words, the intransitive word group “read, arrive at, gain” labeled “reach-destination” has the meaning of “location” or “object” to be reached. reach-destination (b: object) ”, or more precisely“ (reach-destination (o)) (b: object) ”.
A label having this argument is called an event descriptor, and a dictionary (event descriptor dictionary) is created by arranging event descriptors in alphabetical order. Also, the word group for the label is said to be associated with the event descriptor.

In the event descriptor dictionary 13, as shown in FIG. 5, an event descriptor 13B and a physical quantity expression 13C are associated with a predicate 13A indicating an event related to an object.
The physical quantity expression 13C includes a time constraint expression unit 13E including a time constraint expression indicating each section provided by dividing the event in time, and the event in the section in an arbitrary manner such as three-dimensional coordinates, speed, temperature, and humidity. It consists of a set with a physical formula part 13D consisting of physical formulas (P1, P2, P3) expressed in physical quantities.

For example, in FIG. 5A, for each predicate 13A “move, travel, locomote”, an event descriptor 13B “change-location (a, o)”,
P1: do not care (tO ≦ t <t1),
P2: | v |> 0 (t1 ≦ t ≦ t2),
P3: do not care (t2 <t ≦ t3).
Is associated with the physical quantity expression 13C. Here, in the physical quantity expression 13C, t0, t1, t2, and t3 in the time constraint expression unit 13E are parameters that represent real time values, the time t1 corresponds to the occurrence time of the event, and the time t2 is the time. Corresponds to the end time of the event.

Therefore, in the physical formula part 13D, the expression P1 indicates the state of the object in the section immediately before the event occurs, the expression P2 indicates the state of the object in the section in which the event occurs, and the expression P3 indicates the object in the section immediately after the event occurs. Indicates the state of the object.
In the above example, as a condition in the event occurrence period from time t1 to time t2, the condition that the magnitude | v | of the velocity vector v of the object o is larger than 0, that is, the physical movement of the object o. This is defined by the expression part 13D.

In FIG. 5B, an event descriptor 13B “(go-from-region-to-region (o)) (r1: region, r2: region)” and a predicate 13A “move” ,
D (x, r1) = 0 (t0 ≦ t <t1),
| v |> 0 & D (x, r2)> 0 (t1 ≦ t ≦ t2),
D (x, r2) = 0 (t2 <t ≦ t3).
Is associated with the physical quantity expression 13C. Here, in the physical quantity expression 13C, x is the three-dimensional coordinate of the object o, D (r1, r2) represents the Euclidean distance between r1 and r2, and v represents the velocity vector of the object o. In this case, the object o is in the region region1 in the section before the time t1, starts moving from the time t1, reaches the region region2 at the time t2, and expresses that it is in the region 2 in the section after the time t2. . & Indicates a logical product.

In FIG. 5C, for the predicate 13A “drop”, an event descriptor 13B “fall-vertically (o)”;
do not care (tO ≦ t <t1),
(v / | v |) ・ (g / | g |) = 1 & a = 9 (t1 ≦ t ≦ t2),
do not care (t2 <t ≦ t3).
Is associated with the physical quantity expression 13C. Here, in the physical quantity expression 13C, g represents a gravity constant vector, “·” represents an inner product, and a represents the acceleration of the object o. In this case, the object o indicates that the object is moving vertically downward at an acceleration g in a section from time t1 to time t2.
In this way, a physical quantity expression 13C for each event descriptor 13B representing each predicate 13A is formulated and stored in the event descriptor dictionary 13.

[Operation of First Embodiment]
Next, with reference to FIGS. 6 and 7, the operation of the event search apparatus according to the first embodiment of the present invention will be described. FIG. 6 is a flowchart showing event search processing of the event search apparatus according to the first embodiment of the present invention. FIG. 7 is an explanatory diagram showing a specific example of event descriptors and physical quantity expressions.

  In the following, it is assumed that each sensor node 40 (41 to 44) in FIG. 1 includes a three-axis acceleration sensor (see Non-Patent Document 5, etc.), and three axes (XYZ) specific to the acceleration sensor are used as measurement data. Direction) acceleration data and vertical acceleration. The data collection unit 10 calculates the acceleration in the vertical direction and the absolute value of the acceleration in a plane perpendicular to the acceleration based on the measurement data, and sequentially stores them in the sensor DB 11. At this time, in each direction, if the maximum value of the sensor shows a value less than or equal to the threshold value, the acceleration is 0, and if it shows a value greater than or equal to the threshold value, that value is taken as the magnitude of the movement for that time.

The event search device 1 starts the event search process of FIG. 6 according to the operation of the operator. First, the question receiving unit 12 extracts a word indicating an event to be searched from question information input in a natural language by an operator (step 100).
At this time, for example, when question information “vase drop” indicating an event of “dropping a vase” is input, the question reception unit 12 includes a predicate including a verb “drop” indicating the content of the event related to the object, Object information consisting of a noun “vase” indicating the name of the object involved in the event is extracted from the question information.

Next, the question receiving unit 12 searches the event descriptor dictionary 13 for “fall-vertically (o)” as the event descriptor associated with the predicate “drop” (step 101). Here, when the event descriptor and physical quantity expression shown in FIG. 7 are registered in the event descriptor dictionary 13, the physical quantity expression associated with the event descriptor is as follows.
do not care (tO ≦ t <t1); 0 <t1-t0,
(v / | v |) ・ (g / | g |) = 1 & a = g (t1 ≦ t ≦ t2); 200msec ≦ t2-t1 <800msec,
do not care (t2 <t ≦ t3); 0 <t3-t2.
And a search request including the event descriptor “fall-vertically (vase) and physical quantity expression assigned the object information“ vase ”as an argument of the event descriptor is passed to the data search unit 14.

In response to the search request from the question receiving unit 12, the data search unit 14 searches the sensor DB 11 for measurement data corresponding to the event descriptor and physical quantity expression included in the search request (step 102).
Specifically, the range of the value of t2-t1 specified by the physical quantity expression included in the search request is taken out, and since it is between 200 msec and 800 msec (the indoor ceiling is at most 300 cm, free fall from there Since it reaches the floor in less than 800msec at most), it increases in order from 200msec in the meantime, and the acceleration data attached to the measurement data with “vase” included in the event descriptor as object information, that is, “vase” The sensor measurement data is checked from the most recently accumulated data to the past, and the condition “(v / | v |) · (g / | g |) = 1 & a = g” is satisfied. Search for data.

At this time, the initial speed is set to “0 (zero)”, and the speed vector at each time is calculated by numerical integration. The equality determination as to whether or not this condition is satisfied is performed within the error range of the acceleration sensor. For example, when the maximum error of the acceleration sensor is 10%, the equal sign determination is performed in consideration of the error in the range.
Based on the time information of the measurement data obtained as a search result, the data search unit 14 outputs the time corresponding to the time information to the question answering unit 15 as the search result information.

  The question response unit 15 receives the time output as the search result information from the data search unit 14 and answers corresponding to the question information received by the question reception unit 12 from a screen display device (not shown) or a data input / output device. (Step 103). Thereby, a series of event search processing is completed.

  As described above, in the present embodiment, the measurement data measured by the plurality of sensors attached to the object in the environment 3 and the time information indicating the time when the measurement data is measured are obtained by the sensor DB 11. And storing as a set, the event descriptor dictionary 13 uses a word indicating an event related to an object, an event descriptor indicating a group to which the word belongs, and a physical quantity expression of the event expressed by an expression using measurement data, , The predicate indicating the event to be searched is extracted from the question information input in the natural language by the question receiving unit 12, and the event descriptor corresponding to the predicate from the event descriptor dictionary 13 Search for.

Then, the data search unit 14 searches the sensor DB 11 for measurement data corresponding to the physical quantity expression corresponding to the event descriptor obtained by the question reception unit 12, and the search result regarding the event to be searched according to the search result Information is output.
This makes it possible to search for events and phenomena that have occurred in the real world using words and sentences expressed in natural language in an environment in which a large number of sensors that sequentially record changes in the situation are installed.

In the present embodiment, the case where the sensor is attached to an object such as the table 31, the chair 32, the vase 33, the book 34, or the like has been described as an example. However, the present invention is not limited to this. For example, if the person in the room has the RFID tag or the sensor node described above, the name of the person who was in the room when the vase fell from the time information and the record of the RFID tag or sensor node attached to the person Can be searched and displayed.
In addition, if a person does not have an RFID tag or a sensor node, if the room is photographed with a video camera together with the time information, the video at that time can be reproduced instantaneously based on the time information. , Can identify the dropped person instantly.

  In the present embodiment, the question receiving unit 12 extracts a predicate indicating an event to be searched from the question information and target information indicating the target related to the event, and the data search unit 14 performs measurement. Since the measurement data corresponding to the physical quantity expression and the measurement data obtained from the target that matches the target information are searched from the sensor DB 11 when searching the data, the target information is obtained from the question information. By designating, it is possible to search in detail for an event related to a desired object. If no object information is specified, any event that occurs in the environment can be searched regardless of the object.

  In the present embodiment, the question receiving unit 12 searches the event descriptor dictionary 13 for the event descriptor 13B of the event to be searched and its physical quantity expression 13C, and sends these to the data search unit 14 as a search request. The case of passing has been described as an example, but the present invention is not limited to this. For example, the question accepting unit 12 retrieves only the event descriptor 13B and passes it as a retrieval request to the data retrieving unit 14, and the data retrieving unit 14 retrieves the physical quantity expression 13C corresponding to the event descriptor 13B from the event descriptor dictionary 13. You may make it search.

[Second Embodiment]
Next, an event search apparatus according to the second embodiment of the present invention will be described with reference to FIG. FIG. 8 is a block diagram showing a configuration of the event search apparatus according to the second exemplary embodiment of the present invention, and the same or equivalent parts as those in FIG.

  In the first embodiment, a case in which predicate indicating the contents of an event related to an object and object information indicating an object related to the event are specified by the question information input to the question receiving unit 12. Described as an example. In the present embodiment, a case will be described in which question information specifying a predicate modifier that modifies a predicate indicating the content of an event is received by the question receiving unit 12 and a desired event is searched.

  Compared to the first embodiment, an auxiliary descriptor dictionary 16 is added to the event search apparatus 1 according to the present embodiment. Other configurations are the same as those in the first embodiment, and a detailed description thereof is omitted here.

The auxiliary descriptor dictionary 16 is composed of the storage unit described above, and includes a predicate modifier consisting of an adverb, preposition, or adjective indicating a temporal or spatial condition related to the event of the object, and a group to which the predicate modifier belongs. It is a database that stores a correspondence relationship between an auxiliary descriptor including a label to indicate and a physical quantity expression of the event expressed by an expression using measurement data.
FIG. 9 is an explanatory diagram showing a configuration example of the auxiliary descriptor dictionary. Here, “predicate modifiers”, “auxiliary descriptors”, and “physical quantity expressions” indicating events related to the object are stored in association with each other.

[Auxiliary descriptor dictionary]
Next, the auxiliary descriptor dictionary will be described in detail with reference to FIG. FIG. 10 is an explanatory diagram showing a specific example of the auxiliary descriptor dictionary.
The auxiliary descriptor dictionary 16 is generated in advance in the same manner as the event descriptor dictionary 13.

  That is, as auxiliary descriptors based on adverbs with high spatial or temporal appearance frequency, “near (b = object), far (b = object), quickly, slowly, vertically, horizontally” `` At_t (tau), in_t (tau) '' based on related prepositions and `` at_p (b: object), in_p (b = object), on_p (b = object) based on spatial prepositions Is introduced as an auxiliary descriptor. Furthermore, “fast, slow, constant” from the adjective related to “motion” and “low, high, equal ascending, descending” from the adjective related to “scalar physical quantity” are introduced as auxiliary descriptors. A dictionary in which the entire auxiliary descriptor is digitized is created and is called an auxiliary descriptor dictionary.

  Since many events and phenomena occur in time and space, concatenation of event descriptors and auxiliary descriptors enables the description of various events and phenomena. For example, the event descriptor “change-location (person A, chair B)” with a value assigned to an argument and the auxiliary descriptor “in_p (room C)” with a value assigned to an argument “change-location” “location (person A, chair B) in_p (room C)” expresses the event “in room C, person A moves chair B”.

As shown in FIG. 10, in the auxiliary descriptor dictionary 16, the auxiliary descriptor 16B and the physical quantity expression 16C correspond to predicate modifiers 16A such as adverbs, prepositions, and adjectives indicating temporal or spatial conditions related to events. It is attached.
For example, in FIG. 10A, for the predicate modifier 16A adverb “horizontally”, an auxiliary descriptor 16B “horizontally”
P1: do not care (tO ≦ t <t1),
P2: v ・ g = 0 (t1 ≦ t ≦ t2),
P3: do not care (t2 <t ≦ t3).
Is associated with the physical quantity expression 16C. Here, in the physical quantity expression 16C, t0, t1, t2, and t3 in the time constraint expression unit 16E are parameters that represent real time values, the time t1 corresponds to the occurrence time of the event, and the time t2 is the time. Corresponds to the end time of the event.

Therefore, in the physical formula part 16D, the expression P1 indicates the state of the object in the section immediately before the event occurs, the expression P2 indicates the state of the object in the section in which the event occurs, and the expression P3 indicates the object in the section immediately after the event occurs. Indicates the state of the object.
Similarly to the event descriptor, v is the velocity vector of the object o, g is the gravity constant vector, and `` ・ '' indicates the inner product. In the above example, the condition in the event occurrence interval from time t1 to time t2 As a result, the physical formula section 16D defines that the inner product of v and g is zero, that is, the object o is moving horizontally.

In FIG. 10B, an auxiliary descriptor 16B “in_p (b: object)” and a predicate modifier 16A “in”
do not care (tO ≦ t <t1),
D (b, x) = 0 (t1 ≦ t ≦ t2),
do not care (t2 <t ≦ t3).
Is associated with the physical quantity expression 16C. Here, in the physical quantity expression 16C, x represents the three-dimensional coordinates of the object o, D (b, x) represents the Euclidean distance between the objects b and x, and the object o has zero Euclidean distance from the object b. The state, that is, the object o is present in the object b.

In FIG. 10C, an auxiliary descriptor 16B “at_t (tau)” and a predicate modifier 16A “preposition” “at”
do not care (tO ≦ t <tau),
do not care (tau ≦ t ≦ tau),
do not care (tau <t ≦ t3).
Is associated with the physical quantity expression 16C. Here, in the physical quantity expression 16C, tau is a parameter indicating time, and indicates that time tau exists as an arbitrary section provided by dividing an event into time.
In this way, a physical quantity expression 16C for each auxiliary descriptor 16B corresponding to each predicate modifier 16A is formulated and stored in the auxiliary descriptor dictionary 16.

[Operation of Second Embodiment]
Next, with reference to FIG. 11 and FIG. 12, the operation of the event search device according to the second exemplary embodiment of the present invention will be described. FIG. 11 is a flowchart showing event search processing of the event search apparatus according to the second exemplary embodiment of the present invention. FIG. 12 is an explanatory diagram illustrating an example of joining of physical quantity expressions.

  Hereinafter, it is assumed that each sensor node 40 (41 to 44) in FIG. 8 includes a three-axis acceleration sensor (see Non-Patent Document 5, etc.), and three axes (XYZ) specific to the acceleration sensor are used as measurement data. Direction) acceleration data and vertical acceleration. The data collection unit 10 calculates the acceleration in the vertical direction and the absolute value of the acceleration in a plane perpendicular to the acceleration based on the measurement data, and sequentially stores them in the sensor DB 11. At this time, in each direction, if the maximum value of the sensor shows a value less than or equal to the threshold value, the acceleration is 0, and if it shows a value greater than or equal to the threshold value, that value is taken as the magnitude of the movement for that time.

The event search device 1 starts the event search process of FIG. 11 according to the operation of the operator. First, the question receiving unit 12 extracts a word indicating an event to be searched from question information input in a natural language by an operator (step 110).
At this time, for example, when question information “chair move at-tau” indicating an event “moving a chair at time tau” is input, the question reception unit 12 displays “move” indicating the content of the event related to the object. Question information for the predicate modifier "at" that includes the predicate that consists of the predicate that indicates the temporal condition related to the event, and the object information that consists of the noun "chair" that indicates the name of the object involved in the event Extract from

Next, the question receiving unit 12 searches the event descriptor dictionary 13 for “change-location (o)” as the event descriptor 13B associated with the predicate “move” (step 111), and corresponds to this. As physical quantity expression 13C
do not care (tO ≦ t <t1),
| v |> 0 (t1 ≦ t ≦ t2),
do not care (t2 <t ≦ t3).
To get.

In addition, the question receiving unit 12 searches the auxiliary descriptor dictionary 16 for “at_t (tau)” as the auxiliary descriptor 16B associated with the predicate modifier “at” (step 112), and the physical quantity corresponding thereto. As expression 16C
do not care (tO ≦ t <tau),
do not care (tau ≦ t ≦ tau),
do not care (tau <t ≦ t3).
To get.

Thereafter, the question receiving unit 12 joins the event descriptor 13B and the auxiliary descriptor 16B to generate a new event descriptor “change-location (o) at_t (tau)”, and also represents the physical quantity expression 13C and the physical quantity. Join expression 16C. At this time, pattern matching is performed for both time constraint expressions t0 ≦ t <t1, etc., and for the physical expression part, an expression that is the logical product of the respective physical expression parts is obtained. As a result, as shown in FIG.
do not care (tO ≦ t <tau),
| v |> 0 & v ・ g = 0 (tau ≦ t ≦ tau),
do not care (tau <t ≦ t3).
Create a new physical expression part that represents the event moving in the horizontal direction at time tau, and assign the object information "chair" to the argument of this event descriptor as a new event descriptor "change-location A search request including “(o) at_t (tau)” and a new physical quantity expression is passed to the data search unit 14.

In the same manner as described above, in response to the search request from the question receiving unit 12, the data search unit 14 sends measurement data corresponding to the event descriptor, physical quantity expression, and target object information included in the search request from the sensor DB 11. A search is performed (step 113), and based on the time information of the measurement data obtained as a search result, a time corresponding to the time information is output to the question answering unit 15 as search result information.
The question response unit 15 receives the time output as the search result information from the data search unit 14 and answers corresponding to the question information received by the question reception unit 12 from a screen display device (not shown) or a data input / output device. (Step 114). Thereby, a series of event search processing is completed.

  As described above, this embodiment includes predicate modifiers composed of adverbs, prepositions or adjectives indicating temporal or spatial conditions related to events, auxiliary descriptors indicating groups to which these predicate modifiers belong, and measurement data. Is further provided with an auxiliary descriptor dictionary 16 for storing a correspondence relationship between the physical quantity expression of the condition expressed by an expression using, and the question reception unit 12 extracts a predicate and a predicate modifier for the event from the question information. The event descriptor corresponding to the predicate is searched from the event descriptor dictionary 13 and the auxiliary descriptor corresponding to the predicate modifier is searched from the auxiliary descriptor dictionary 16.

Then, the data search unit 14 searches the sensor DB 11 for measurement data corresponding to the physical quantity expression corresponding to the event descriptor and the auxiliary descriptor obtained by the question reception unit 12, and becomes a search target according to the search result. The search result information about the event is output.
Thus, an event can be searched for using a predicate indicating the event and a predicate modifier that modifies the predicate, and a desired event can be searched under more detailed conditions.

  In the present embodiment, the question receiving unit 12 joins the event descriptor 13B and the auxiliary descriptor 16B, joins the physical quantity expression 13C and the physical quantity expression 16C, and then uses them as a search request. Although the case where it passes to 14 was demonstrated as an example, it is not limited to this. For example, from the question receiving unit 12, the event descriptor 13B, the auxiliary descriptor 16B, the physical quantity expression 13C, and the physical quantity expression 16C are directly passed to the data search unit 14 as a search request without being joined. While joining the child 13B and the auxiliary descriptor 16B and joining the physical quantity expression 13C and the physical quantity expression 16C, the corresponding measurement data may be searched.

  Alternatively, the event receiving unit 12 passes the event descriptor 13B and the auxiliary descriptor 16B together as a search request to the data search unit 14 without joining them, and the data search unit 14 joins the event descriptor 13B and the auxiliary descriptor 16B. At the same time, the physical quantity expression 13C and the physical quantity expression 16C corresponding to the event descriptor 13B and the auxiliary descriptor 16B are retrieved from the event descriptor dictionary 13 and the auxiliary descriptor dictionary 16, respectively, and the physical quantity expression 13C and the physical quantity expression 16C are retrieved. After joining, corresponding measurement data may be searched.

[Third Embodiment]
Next, an event search apparatus according to the third embodiment of the present invention will be described with reference to FIG. FIG. 13 is a block diagram showing a configuration of an event search apparatus according to the third exemplary embodiment of the present invention. The same reference numerals are given to the same or equivalent parts as those in FIG.

  In the first embodiment, when the question information input to the question receiving unit 12 includes a predicate indicating the content of an event related to the object, and further object information indicating the object related to the event is added. Described as an example. In the present embodiment, a case will be described in which question information that is not a word as described above but a sentence such as a question sentence is received by the question receiving unit 12 to search for a desired event.

Compared to the first embodiment, a syntax analysis unit 17 is added to the event search apparatus 1 according to the present embodiment. Other configurations are the same as those in the first embodiment, and a detailed description thereof is omitted here.
The syntax analysis unit 17 includes the above-described dedicated arithmetic processing circuit unit and arithmetic processing unit, and has a function of analyzing the syntax of the question information received by the question receiving unit 12. A known technique may be applied to the syntax analysis processing in the syntax analysis unit 17, and detailed description thereof is omitted here.

[Operation of Third Embodiment]
Next, the operation of the event search apparatus according to the third exemplary embodiment of the present invention will be described with reference to FIG. FIG. 14 is a flowchart showing event search processing of the event search apparatus according to the third embodiment of the present invention. FIG. 15 is an explanatory diagram showing a specific example of event descriptors, auxiliary descriptors, and physical quantity expressions.

  Hereinafter, it is assumed that each sensor node 40 (41 to 44) in FIG. 13 includes a three-axis acceleration sensor (see Non-Patent Document 5, etc.), and three axes (XYZ) specific to the acceleration sensor are used as measurement data. Direction) acceleration data and vertical acceleration. The data collection unit 10 calculates the acceleration in the vertical direction and the absolute value of the acceleration in a plane perpendicular to the acceleration based on the measurement data, and sequentially stores them in the sensor DB 11. At this time, in each direction, if the maximum value of the sensor shows a value less than or equal to the threshold value, the acceleration is 0, and if it shows a value greater than or equal to the threshold value, that value is taken as the magnitude of the movement for that time.

The event search device 1 starts the event search process of FIG. 14 according to the operation of the operator. First, the question receiving unit 12 receives the question information input in natural language by the operator, and instructs the syntax analysis unit 17 to perform syntax analysis.
The syntax analysis unit 17 parses the question information in accordance with an instruction from the question reception unit 12, and returns the result to the question reception unit 12 (step 120).
The question reception unit 12 extracts a predicate and a predicate modifier that indicate an event to be searched based on the analysis result from the syntax analysis unit 17 (step 121).

For example, a case is assumed in which a person takes out the book 34 from the bookshelf, places it on the table 31, and leaves the room without performing the act of returning it to the original bookshelf. At that time, in order to know when and who has taken out the book 34 from the bookshelf and left it on the table 31, the question information “When did a book moved from a bookshelf to a table?” Is input.
The parsing unit 17 performs parsing on the question information, and “a book” is the subject of “moved”, and the two preposition phrases “from a bookshelf” and “to a table” depend on “moved”. That “moved” is the past tense of “move”.

  From such an analysis result, the question receiving unit 12 extracts a predicate “move” indicating the contents of the event related to the object and object information “a book” indicating the name of the object related to the event. .

Next, the question reception unit 12 sets “go-from-region-to-region (o)) (r1, r2)” as the event descriptor associated with the predicate “move” from the event descriptor dictionary 13. Search is performed (step 122). Here, when the event descriptor and physical quantity expression shown in FIG. 15 are registered in the event descriptor dictionary 13, the physical quantity expression associated with the event descriptor is as follows.
D (x, r1) = 0 (tO ≦ t <t1); t1-t0 = 200msec,
| v |> 0 & D (x, r2)> 0 (t1 ≦ t ≦ t2); 200msec ≦ t2-t1 <10000msec,
D (x, r2) = 0 (t2 <t ≦ t3); t3-t2 = 200msec.
Event descriptor `` go-from-region-to-region (book) '' with object information `` book '', `` bookshelf '', and `` table '' assigned to argument x, r1, r2 of this event descriptor ) (bookshelf, table) ”and the physical quantity representation are passed to the data search unit 14.

In response to the search request from the question receiving unit 12, the data search unit 14 searches the sensor DB 11 for measurement data corresponding to the event descriptor and physical quantity expression included in the search request (step 123).
Specifically, first, in order to check the second condition | v |> 0 & D (x, r2)> 0 (t1 ≦ t ≦ t2), the range of the value of t2−t1 is 200 msec or more and 10 seconds or less. Position data based on the ultrasonic beacon attached to the book, increasing t2-t1 in order from 200 msec in the range of (maximum time to move from end to end in the room corresponding to 10 seconds here) The sensor DB 11 is checked from the most recently accumulated to the oldest, and the velocity vector is calculated from the position data.

  Then, measurement data whose absolute value is greater than 0 and whose distance from the table calculated based on the position data is greater than 0 is detected. Further, before and after the measurement data, here, t1-t0 and t3-t2 are 200 msec, the first condition D (x, r1) = 0 (tO ≦ t <t1) and the third condition D (x, r2 ) = 0 (t2 <t ≦ t3) is checked to see if it is satisfied. At this time, the equality determination as to whether or not each condition is satisfied is performed within the range of error of the position detection sensor.

Based on the time information of the measurement data obtained as a search result, the data search unit 14 outputs the time corresponding to the time information to the question answering unit 15 as the search result information.
The question response unit 15 receives the time output as the search result information from the data search unit 14, and at this time, since the question sentence is a question regarding When and Time, the measurement data satisfying both of the above three conditions is received. Based on the time information, an answer “XX year YY month ZZ day AA hour BB minute CC second” is output from a screen display device (not shown) or a data input / output device (step 124). Thereby, a series of event search processing is completed.

  As described above, the present embodiment further includes the syntax analysis unit 17 for analyzing the syntax of the question information, and the question receiving unit 12 extracts various words indicating events based on the analysis result in the syntax analysis unit 17. As a result, it is possible to search for any event using question information consisting of sentences such as question sentences instead of enumeration of words consisting of natural language, and even general users can easily use the event search device. It becomes possible.

  In the present embodiment, the case is described in which time information when a desired event occurs is output as a search result. However, the present invention is not limited to this. For example, if the question text is Who, if the RFID tag is attached to the person, the name of the person who was in the room when the book moved based on the information may be output, and the person who handled the book Can be output as search results. If no RFID tag is attached to a person, for example, if the room is photographed with a video camera together with time information, the video at that time can be instantaneously reproduced based on the time information. The person who handled the book can be identified instantly.

[Fourth Embodiment]
Next, an event search apparatus according to the fourth embodiment of the present invention will be described with reference to FIG. FIG. 16 is a block diagram showing the configuration of the event search apparatus according to the fourth embodiment of the present invention, and the same or equivalent parts as those in FIG.

  In the first embodiment, when the measurement data related to a desired event is searched from the sensor DB 11, the sensor DB 11 is searched based on the event descriptor corresponding to the predicate or object information specified by the question information and the physical quantity expression. Explained when to do. In the present embodiment, a case will be described in which measurement data relating to a desired event is retrieved from the sensor DB 11 using a data pattern corresponding to the event descriptor.

  Compared to the first embodiment, an event reference database (hereinafter referred to as an event reference DB) 18 is added to the event search apparatus 1 according to the present embodiment. Other configurations are the same as those in the first embodiment, and a detailed description thereof is omitted here.

The event reference DB 18 is a database that includes the storage unit described above, and stores a data pattern indicating a change in measurement data when the event occurs for each event descriptor corresponding to a predicate indicating the event of the target object. .
FIG. 17 is an explanatory diagram of a configuration example of the event reference database. Here, an “event descriptor” indicating an event related to the object is associated with a “data pattern”. The data pattern may be generated by measuring measurement data in advance through experiments and calculating typical data using various statistical processes.

[Operation of Fourth Embodiment]
Next, the operation of the event search apparatus according to the fourth exemplary embodiment of the present invention will be described with reference to FIG. FIG. 18 is a flowchart showing event search processing of the event search apparatus according to the fourth embodiment of the present invention.

  Hereinafter, it is assumed that each sensor node 40 (41 to 44) in FIG. 16 includes a three-axis acceleration sensor (see Non-Patent Document 5, etc.), and three axes (XYZ) specific to the acceleration sensor are used as measurement data. Direction) acceleration data and vertical acceleration. The data collection unit 10 calculates the acceleration in the vertical direction and the absolute value of the acceleration in a plane perpendicular to the acceleration based on the measurement data, and sequentially stores them in the sensor DB 11. At this time, in each direction, if the maximum value of the sensor shows a value less than or equal to the threshold value, the acceleration is 0, and if it shows a value greater than or equal to the threshold value, that value is taken as the magnitude of the movement for that time.

The event search device 1 starts the event search process of FIG. 18 according to the operation of the operator. First, the question receiving unit 12 extracts a word indicating an event to be searched from question information input in a natural language by an operator (step 130).
At this time, for example, when question information “vase drop” indicating an event of “dropping a vase” is input, the question reception unit 12 includes a predicate including a verb “drop” indicating the content of the event related to the object, Object information consisting of a noun “vase” indicating the name of the object involved in the event is extracted from the question information.

Next, the question receiving unit 12 searches the event descriptor dictionary 13 for “fall-vertically (o)” as the event descriptor associated with the predicate “drop” (step 131). Here, when the event descriptor and physical quantity expression shown in FIG. 7 are registered in the event descriptor dictionary 13, the physical quantity expression associated with the event descriptor is as follows.
do not care (tO ≦ t <t1); 0 <t1-t0,
(v / | v |) ・ (g / | g |) = 1 & a = g (t1 ≦ t ≦ t2); 200msec ≦ t2-t1 <800msec,
do not care (t2 <t ≦ t3); 0 <t3-t2.
And a search request including the event descriptor “fall-vertically (vase) and physical quantity expression assigned the object information“ vase ”as an argument of the event descriptor is passed to the data search unit 14.

  In response to the search request from the question receiving unit 12, the data search unit 14 searches the event reference DB 18 for a data pattern corresponding to the event descriptor included in the search request (step 132), and the event included in the search request The measurement data corresponding to the descriptor and physical quantity expression and the most similar to the data pattern is searched from the sensor DB 11 (step 133).

  Specifically, the range of the value of t2-t1 specified by the physical quantity expression included in the search request is taken out, and since it is between 200 msec and 800 msec (the indoor ceiling is at most 300 cm, free fall from there Since it reaches the floor in less than 800msec at most), it increases in order from 200msec in the meantime, and the acceleration data attached to the measurement data with “vase” included in the event descriptor as object information, that is, “vase” The sensor DB 11 is checked from the latest accumulated measurement data to the past, and the distance from the data pattern obtained from the event reference DB 18 is sequentially calculated.

At this time, the distance between the measurement data and the data pattern may be calculated using, for example, time expansion / contraction dynamic programming (see Non-Patent Document 6), or another matching degree such as a correlation value may be used as the distance.
In this way, the distance is calculated sequentially, some measurement data is selected in ascending order of the distance, and the time corresponding to the time information of the measurement data is output to the question answering unit 15 as the search result information.

  The question response unit 15 receives the time output as the search result information from the data search unit 14 and answers corresponding to the question information received by the question reception unit 12 from a screen display device (not shown) or a data input / output device. (Step 134). Thereby, a series of event search processing is completed.

  As described above, in the present embodiment, when the event reference DB 18 stores a data pattern indicating a change in measurement data related to an object for each event descriptor, and when the measurement data is searched by the data search unit, Since the data pattern corresponding to the event descriptor obtained by the question receiving unit is searched from the event reference DB 18 and the measurement data changing in a similar manner to the data pattern is searched from the sensor DB 11, the physical quantity expression is used. Compared to the case of searching measurement data, the search processing load can be reduced, and a desired event can be searched in a shorter time.

[Fifth Embodiment]
Next, an event search device according to a fifth embodiment of the present invention will be described. Note that the block diagram showing the configuration of the event search apparatus according to the present embodiment is the same as FIG. 16 described above, and will be described with reference to FIG.

  In the fourth embodiment, a case has been described in which measurement data relating to a desired event is retrieved from the sensor DB 11 using a data pattern corresponding to the event descriptor. In the present embodiment, a case will be described in which measurement data related to a desired event is searched from the sensor DB 11 using segment information that specifies measurement data corresponding to an event descriptor.

  Similar to the fourth embodiment, an event reference database (hereinafter referred to as an event reference DB) 18 is added to the event search apparatus 1 according to the present embodiment as compared with the first embodiment. Yes. Other configurations are the same as those in the first embodiment, and a detailed description thereof is omitted here.

The event reference DB 18 is composed of the storage unit described above, and stores segment information including a storage address and management information for specifying measurement data related to the event for each event descriptor corresponding to the predicate indicating the event of the target object. It is a database.
FIG. 19 is an explanatory diagram of a configuration example of the event reference database. Here, “event descriptor” indicating an event related to the object is associated with “segment information”. For the segment information, the correspondence relationship between the event and the measurement data is examined in advance, and the storage address and management information for specifying the measurement data corresponding to each event may be stored in the event reference DB 18 as segment information.

[Operation of Fifth Embodiment]
Next, with reference to FIG. 20, the operation of the event search device according to the fifth exemplary embodiment of the present invention will be described. FIG. 20 is a flowchart showing event search processing of the event search apparatus according to the fifth embodiment of the present invention.

  Hereinafter, it is assumed that each sensor node 40 (41 to 44) in FIG. 16 includes a three-axis acceleration sensor (see Non-Patent Document 5, etc.), and three axes (XYZ) specific to the acceleration sensor are used as measurement data. Direction) acceleration data and vertical acceleration. The data collection unit 10 calculates the acceleration in the vertical direction and the absolute value of the acceleration in a plane perpendicular to the acceleration based on the measurement data, and sequentially stores them in the sensor DB 11. At this time, in each direction, if the maximum value of the sensor shows a value less than or equal to the threshold value, the acceleration is 0, and if it shows a value greater than or equal to the threshold value, that value is taken as the magnitude of the movement for that time.

The event search device 1 starts the event search process of FIG. 20 according to the operation of the operator. First, the question receiving unit 12 extracts a word indicating an event to be searched from question information input in a natural language by an operator (step 140).
At this time, for example, when question information “vase drop” indicating an event of “dropping a vase” is input, the question reception unit 12 includes a predicate including a verb “drop” indicating the content of the event related to the object, Object information consisting of a noun “vase” indicating the name of the object involved in the event is extracted from the question information.

Next, the question receiving unit 12 searches the event descriptor dictionary 13 for “fall-vertically (o)” as the event descriptor associated with the predicate “drop” (step 141). Here, when the event descriptor and physical quantity expression shown in FIG. 7 are registered in the event descriptor dictionary 13, the physical quantity expression associated with the event descriptor is as follows.
do not care (tO ≦ t <t1); 0 <t1-t0,
(v / | v |) ・ (g / | g |) = 1 & a = g (t1 ≦ t ≦ t2); 200msec ≦ t2-t1 <800msec,
do not care (t2 <t ≦ t3); 0 <t3-t2.
And a search request including the event descriptor “fall-vertically (vase) and physical quantity expression assigned the object information“ vase ”as an argument of the event descriptor is passed to the data search unit 14.

  In response to the search request from the question receiving unit 12, the data search unit 14 searches the event reference DB 18 for segment information corresponding to the event descriptor included in the search request (step 142), and the event included in the search request The measurement data corresponding to the descriptor and the physical quantity representation and specified by the segment information is searched from the sensor DB 11 (step 143).

  Specifically, the range of the value of t2-t1 specified by the physical quantity expression included in the search request is taken out, and since it is between 200 msec and 800 msec (the indoor ceiling is at most 300 cm, free fall from there Since it reaches the floor in a time shorter than 800 msec at most), it increases in order from 200 msec in the meantime, and the measurement data with “vase” included in the event descriptor as object information, that is, the acceleration attached to “vase” For the measurement data specified by the segment information, the sensor DB 11 is checked from the latest accumulated data to the past.

In this way, the measurement data corresponding to the event descriptor, physical quantity expression, and object information included in the search request is searched from the sensor DB 11, and based on the time information of the measurement data obtained as a search result, The time corresponding to the time information is output to the question answering unit 15 as search result information.
The question response unit 15 receives the time output as the search result information from the data search unit 14 and answers corresponding to the question information received by the question reception unit 12 from a screen display device (not shown) or a data input / output device. (Step 144). Thereby, a series of event search processing is completed.

  As described above, in the present embodiment, when the event reference DB 18 stores segment information for specifying measurement data related to an object for each event descriptor, and the data search unit 14 searches for measurement data, The segment information corresponding to the event descriptor obtained by the question receiving unit 12 is searched from the event reference DB 18, and the measurement data corresponding to the physical quantity expression corresponding to the event descriptor is measured from the measurement data corresponding to the segment information from the sensor DB 11. Compared to searching for measurement data that corresponds to the physical quantity expression for all measurement data, it is possible to narrow down the measurement data to be searched, and to find the desired event in a shorter time. It becomes possible to search.

[Extended embodiment]
In each of the above embodiments, the case where the time information indicating the time at which the event occurred is output as a search result for the question information based on the measurement time of the measurement data searched from the sensor DB 11 has been described as an example. The content of the search result to be output is not limited to this. For example, as shown in FIG. 3, when object information including the name and identification information of an object is attached to each measurement data as the sensor DB 11, the object information may be output as a search result. Good. When the measurement data indicates coordinate information (position information), a place name corresponding to the coordinate information may be output as the place where the event has occurred. In addition, you may output the name regarding the person relevant to the event, the other target object, and the image | video of the time when the said event generate | occur | produced.

It is a block diagram which shows the structure of the event search device concerning the 1st Embodiment of this invention. It is explanatory drawing which shows the structural example of measurement data. It is explanatory drawing which shows the structural example of a sensor database. It is explanatory drawing which shows the structural example of an event descriptor dictionary. It is explanatory drawing which shows the specific example of an event descriptor dictionary. It is a flowchart which shows the event search process of the event search device concerning the 1st Embodiment of this invention. It is explanatory drawing which shows the specific example of an event descriptor and physical quantity expression. It is a block diagram which shows the structure of the event search device concerning the 2nd Embodiment of this invention. It is explanatory drawing which shows the structural example of an auxiliary descriptor dictionary. It is explanatory drawing which shows the specific example of an auxiliary descriptor dictionary. It is a flowchart which shows the event search process of the event search device concerning the 2nd Embodiment of this invention. It is explanatory drawing which shows the example of joining of physical quantity expression. It is a block diagram which shows the structure of the event search device concerning the 3rd Embodiment of this invention. It is a flowchart which shows the event search process of the event search device concerning the 3rd Embodiment of this invention. It is explanatory drawing which shows the specific example of an event descriptor, an auxiliary descriptor, and physical quantity expression. It is a block diagram which shows the structure of the event search device concerning the 4th Embodiment of this invention. It is explanatory drawing which shows the structural example of an event reference database. It is a flowchart which shows the event search process of the event search device concerning the 4th Embodiment of this invention. It is explanatory drawing which shows the structural example of an event reference database. It is a flowchart which shows the event search process of the event search device concerning the 5th Embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Event search apparatus, 10 ... Data collection part, 11 ... Sensor DB, 12 ... Question reception part, 13 ... Event descriptor dictionary, 13A ... Predicate, 13B ... Event descriptor, 13C ... Physical quantity expression, 13D ... Physical formula part , 13E ... time constraint formula part, 14 ... data search part, 15 ... question answering part, 16 ... auxiliary descriptor dictionary, 16A ... predicate modifier, 16B ... auxiliary descriptor, 16C ... physical quantity expression, 17 ... syntax analysis part, 18 ... event reference DB, 2 ... sensor network, 3 ... environment, 31 ... table, 32 ... chair, 33 ... vase, 34 ... book, 40, 41-44 ... sensor node.

Claims (9)

An event search device that searches for any event that occurs in the environment by searching a sensor database that stores measurement data measured by a plurality of sensors attached to objects in the environment,
A sensor database for storing measurement data measured by a plurality of sensors attached to an object in the environment and time information indicating the time at which the measurement data was measured, as a set;
An event description storing a correspondence relationship between a predicate consisting of a verb indicating an event related to the object, an event descriptor indicating a group to which the predicate belongs, and a physical quantity expression of the event expressed by an expression using the measurement data Child dictionaries,
A question accepting unit that extracts a predicate indicating an event to be searched from question information input in a natural language, and searches for an event descriptor corresponding to a previous description word from the event descriptor dictionary;
The measurement data corresponding to the physical quantity expression corresponding to the event descriptor obtained by the question receiving unit is searched from the sensor database, and search result information regarding the event to be searched is output according to the search result An event search device comprising: a data search unit that performs: In the event search device according to claim 1,
The question reception unit extracts object information indicating an object related to an event to be searched from the question information,
The data search unit, when searching the measurement data, searches the sensor database for measurement data corresponding to the physical quantity expression and obtained from an object that matches the object information. An event search device characterized by the above. In the event search device according to claim 1,
Predicate modifiers consisting of adverbs, prepositions, or adjectives indicating temporal or spatial conditions related to the event, auxiliary descriptors indicating the groups to which these predicate modifiers belong, and expressions using the measurement data An auxiliary descriptor dictionary for storing a correspondence relationship between the physical quantity representation of the condition and
The question receiving unit extracts a predicate modifier consisting of an adverb, preposition, or adjective indicating a temporal or spatial condition related to the event from the question information, and converts the predicate modifier from the auxiliary descriptor dictionary Search for the corresponding auxiliary descriptor,
When searching for the measurement data, the data search unit searches the sensor database for measurement data corresponding to each physical quantity expression corresponding to the combined descriptor and the auxiliary descriptor obtained by the question receiving unit. An event search device characterized by the above. In the event search device according to claim 1,
The physical quantity expression includes a time constraint expression part including a time constraint expression indicating each section before, during, and after the occurrence of the event, and a physical expression including a physical expression expressing the event in the section with an arbitrary physical quantity. An event search device characterized by comprising a set of parts. In the event search device according to claim 1,
A parsing unit for analyzing the syntax of the question information;
The event search device, wherein the question receiving unit extracts a previous description word based on an analysis result in the syntax analysis unit. In the event search device according to claim 1,
An event reference database for storing a data pattern indicating a change in measurement data related to the object for each event descriptor;
When searching for the measurement data, the data search unit searches the event reference database for a data pattern corresponding to the event descriptor obtained by the question reception unit, and changes in a similar manner to the data pattern. An event search device, wherein the measurement data to be searched is searched from the sensor database. In the event search device according to claim 1,
An event reference database storing segment information for specifying measurement data related to the event for each event descriptor;
When searching for the measurement data, the data search unit searches the event reference database for segment information corresponding to the event descriptor obtained by the question reception unit, and measures data corresponding to the segment information An event search device, wherein measurement data corresponding to a physical quantity expression corresponding to the event descriptor is searched from. An event search method used in an event search apparatus for searching for an arbitrary event occurring in the environment by searching a sensor database that stores measurement data measured by a plurality of sensors attached to an object in the environment. There,
Storing, as a set, measurement data measured by a plurality of sensors attached to an object in the environment and time information indicating the time at which the measurement data was measured by the sensor database;
Correspondence between a predicate consisting of a verb indicating an event related to the object, an event descriptor indicating a group to which the predicate belongs, and a physical quantity expression of the event expressed by an expression using the measurement data by the event descriptor dictionary Memorizing the relationship;
Extracting a predicate indicating an event to be searched from the question information input in natural language by the question receiving unit, and searching for an event descriptor corresponding to the previous description word from the event descriptor dictionary;
The data search unit searches the sensor database for measurement data corresponding to the physical quantity expression corresponding to the event descriptor obtained by the question reception unit, and relates to the event to be searched according to the search result An event search method comprising: outputting search result information. A computer of an event search device that searches for any event that occurs in the environment by searching a sensor database that stores measurement data measured by a plurality of sensors attached to objects in the environment,
Storing, as a set, measurement data measured by a plurality of sensors attached to an object in the environment and time information indicating the time at which the measurement data was measured by the sensor database;
Correspondence between a predicate consisting of a verb indicating an event related to the object, an event descriptor indicating a group to which the predicate belongs, and a physical quantity expression of the event expressed by an expression using the measurement data by the event descriptor dictionary Memorizing the relationship;
Extracting a predicate indicating an event to be searched from the question information input in natural language by the question receiving unit, and searching for an event descriptor corresponding to the previous description word from the event descriptor dictionary;
The data search unit searches the sensor database for measurement data corresponding to the physical quantity expression corresponding to the event descriptor obtained by the question reception unit, and relates to the event to be searched according to the search result A program that executes a step of outputting search result information.

Images