In this preliminary report, we show how action planning is realized by using LOD datasets, e.g., Linked Geo Data, DBpedia, WordNet, etc. To make a recommendation for car drivers and passengers, we combine these existing open datasets with newly constructed ontologies of facilities and services. We develop the inference procedure to translate user requests into SPARQL queries to obtain a recommendation on appropriate facilities for users. Common sense knowledge is also required in the reasoning process.
While Linked Data is now gradually growing to be the infrastructure of coming Knowledge Society, we are still struggling to show the potential of Linked Data to most people in the society including basic industries. To cope with this situation and propel the deployment of Semantic Web technology, it is needed to demonstrate the performance of linking distinct datasets and show the usefulness of outbound and inbound linking data beyond enterprise data in diverse applications. Yet there is no linking data among large linked datasets such as DBpedia, Freebase, and OpenCyc from the viewpoint of LOD applications, although each collection of them are a kind of isolated showcase of LOD with internally linked data within their own territories and objectives.
In our use-case, the system accepts ambiguous requests from car drivers and passengers, plans driver actions to achieve goals that satis es the requests, including alternatives, and makes a recommendation for the drivers and passengers.
In this preliminary work, we have found that it is required more goal-oriented linked datasets and common sense knowledge as bridge between isolated LOD datasets existing. We have also found that Semantic Web technology or specifically RDF stores and SPARQL engines are enough as enabling technology to create and demonstrate new applications based on heterogeneous and diverse datasets.
To obtain driving destinations as goal, we arranged Linked Geo Data and
DBpedia Japanese[
The purpose of this preliminary report is to make a clear direction for development of LOD applications in order to deploy linked data as the infrastructure of society in future. The structure of this preliminary paper is as follows. We describe the detail of the use-case in Section 2. Section 3 reports the related work from the viewpoint of the long-term research activity on arti cial intelligence. In Section 4, we show how we realize new ontologies on facilities and services in order to utilize open knowledge-bases and LOD. Section 5 describes the systematization of multiple linked datasets and the SPARQL endpoint. Section 6 describes the inference procedure for the purpose of getting recommendations by using SPARQL queries. We show how we can realize action planning by using open knowledge-bases and LOD. Section 7 reports a simple example of execution by this prototype system that realizes a new purpose-oriented inference engine with SPARQL and large-scale open knowledge-bases. Section 8 provides discussions. In the last section, we summarize the results and address the future work toward the new era of Knowledge Society based on Open Data and Linked Data. 2
In setting of the use-case, we rstly made more than ten scenarios of conversation between users and this system. In each case, a user in a car speaks a single or a number of requests to do something with driving a car. Then, the system analyzes the requests under the consideration of current contexts such as time, location, driving time, etc. At last, the system makes concrete action proposals to visit speci c points (shop, facility, etc.) or areas (sightseeing area, good place for time-consuming, etc.) with a reasonable visiting order. Basically, the request may be vague and complex, but the recommendation is speci c and concrete. However, every recommendation is a sequence of actions, and proposed actions are quite limited within these scenarios, for example, drive somewhere, buy or eat something, do some sport, and so on. One of the simplest scenarios is as follows.
Child passenger(hereafter C): I want to see a lion.
System(hereafter S): How about Ueno Zoo. A baby lion was born recently.
C: It sounds good, but I was there last month.
S: Well, how about Kinoshita Circus. You can see a lion show there. C: OK. That's ne.
In this scenario, the system must discover the knowledge that a lion is a kind of animal and a zoo is a public entertainment facility for seeing animals. The system must nd out a nearest zoo, that is Ueno Zoo in this case, from the current location, and must reason that users have enough time to drive to the destination and walking around the zoo. Furthermore, due to the negative response of the user, the system must discover a neighboring circus that presents a lion show as an alternative. 3
Planning was one of the most popular AI research area in the 1970s through the
1980s, where the research efforts focused on reasoning mechanisms on making
plans [
The recent success of IBM Watson in TV show Jeopardy! seems to promise
knowledge graph approach for problem solving and decision making[
To endow computers with common sense is one of the major long-term goals
of arti cial intelligence. Common sense reasoning widely ranges over a number
of different elds from taxonomic reasoning, geographic reasoning, temporal
reasoning, reasoning about actions and changes, qualitative reasoning [
The role of verb is not seriously regarded in action planning so far. Action is just called operator in the context of old AI planning. Cognitive Linguistics pays more careful attention on the relation between verbs and objectives. In this research work, we picked up several verbs such as `see', `eat', and `buy' in order to plan actions according to the use-case scenarios, and the relation of such verbs to objectives is realized in our facility ontology and service ontology. See the details in the following section.
Schank addressed eleven primitive actions in Conceptual Dependency
theory [
Frame theory by Fillmore is a theory for Natural Language
Understanding [
Levin [
Generally, we have a number of aspects in dialogue. Searle described the
mechanism of human speech interaction and addressed the Speech Act theory
[
Instead of directly searching individual facilities like Ueno Zoo or individual shops like Yodobashi Akiba store (a home electric appliance mass retailer in Japan), we considered classes of facilities like zoo or home electric appliance mass retailer to make the system scalable, then made a facility ontology that contains typical facilities and we de ned typical users' behavior at such facilities like \a user sees animals in a zoo" or \a user buys a household appliance at a home electric appliance mass retailer." Even if we accidentally fail to guide an actual facility that satis es user's special requests, such a problem will be solved with the development of richer and more speci c datasets that includes individual facilities.
The facility ontology is constructed mainly by extracting facility classes related to leisure and meals in Linked Geo Data (LGD). LGD constructs a shallow class hierarchy from tags attached to the nodes and ways of OpenStreetMap (OSM). Therefore, LGD classes makes it easy to incorporate new facilities and new facility types.
On the other hand, as a result of adopting LGD / OSM, duplicates of classes due to notation uctuation of tags and the low coverage rate of actual facilities at the instance level could be a big problem. However, we think this approach is the best for our purpose in our best knowledge, because the LGD / OSM is the largest facility data that is freely available at the present. Also note that actually it is impossible to measure how much the existing facilities are covered in reality. Regarding duplicates of classes in LGD, we select an entity as primary class that has both the most information-rich descriptions on the OSM and a large number of instances, then the rest are associated with owl:equivalentClass to the primary class.
The following shows an example of zoo class in the facility ontology. The meanings of Japanese words are added here in English as turtle comments for readers. Both a service of \see animal" and \pay admission fee for cultural facility" are actually described in the service ontology as subclasses of \see" service and \admission-viewing-gaming" service. Note that each service is described as a pair of an action and an action target, which users can perform. In this paper, we manually acquired and created service knowledge of facilities within the scenarios as necessary. See the statistic numbers in Table 1. As shown below, the lgdo:Zoo class is linked to the dbo:Zoo class in DBpedia Ontology to make possible to search related facility instances in DBpedia Japanese. The dbo:Zoo already has a link to Wikidata's wikidata:Q43501. Thus, it can be easily expanded when Wikidata is added. lgdo:Zoo a owl:Class; servicevoc:dbpediaClass dbo:Zoo ; servicevoc:provideService [ servicevoc:hasService [ servicevoc:action action: ; servicevoc:target target: servicevoc:action action: ; servicevoc:target target: ]] ; rdfs:subClassOf servicevoc:Facility . 4.2
In the facility ontology, a number of services corresponding to distinct facilities come up with common abstract services. For example, both museums and art museums have the same service of \paying entrance fee for cultural facilities". In addition, there are hierarchical relationships among users' action targets, then we have a similar relationship between services. For example, \seeing animals" can be regarded as the top of \looking at a lion". We constructed an ontology of # pay ], [ # admission fee # for cultural facility # see # animal services apart from facility classes, so that services are independently recognizable, and it enabled us to expand the performance of inference by applying the hierarchy of services. In this paper, the part of service ontology is constructed by using the Classi cation in the Household Survey of the Ministry of Internal Affairs and Communications. The top of service ontology is the `facility service' and it is related to aspects of two types of behaviors, namely, `purchase service' focused on purchasing behavior, and an `activity service' focused on the other behaviors at facilities. The following shows an example of `purchase service' ontology entries.
service: _ a owl:Class; rdfs:label " _ "; servicevoc:action action: ; servicevoc:target target: ; rdfs:subClassOf service: _ service: _ a owl:Class; rdfs:label " _ "; servicevoc:action action: ; servicevoc:target target: ; rdfs:subClassOf service: _ # food service # buy # food . # purchase service # meat service # buy # meat . # food service 4.3
For the sake of systematical description of actions and action targets, we used the Household Income Balance Item Classi cation List (January, 2015) of the Statistics Bureau of the Ministry of Internal Affairs and Communications, of which items of statistics data are used to describe purchasing behavior at facilities. User's behavior at facilities can be divided into purchasing behavior (such as buying something or paying for some bene ts as service) and the other actions (see, eat, drink, etc.). This classi cation is based on a hierarchical structure of action targets as users' behavior as consumer, so it is possible to consider cooperation with statistical data in future, starting with purchase actions. For actions and action targets other than purchasing behavior, we used Japanese WordNet, because we want to use WordNet's knowledge on the relationship between each verb as action and each noun as an action target. For instance, we made Action Target Ontology as follows. target: rdfs:label " "; servicevoc:wordnet wnja11instances:word# animal . target: a owl:Class; rdfs:label " "; servicevoc:wordnet wnja11instances:wordrdfs:subClassOf target: . All LOD datasets and ontologies in this study and their statistics data are described in Table 1.
The service ontology at the bottom of the table is the ontology we constructed this time, as explained in the above.
While the number of data with longitude and latitude were respectively 26,351,904, 100,139, and 1,014,836 for LinkedGeoData, DBpedia Japanese, and DBpedia respectively, the number of geodata, of which each is close to rectangle, in domestic portion excluding Hokkaido, is respectively 538,878, 67,199, and 15,409. 5
We have collected a number of open knowledge resources as shown at the upper part of Table 1, and all of them are stored in one RDF store. However, at the time of this writing, we have actually used only DBpedia Japanese, LinkedGeoData, Japanese WordNet, and DBpedia Ontology as open datasets. Wikidata is not stored because of the capacity.
The system used one endpoint built with one dedicated RDF store.
In this preliminary research, we process natural sentences only within the range expected at use-cases. Furthermore, in this paper it is assumed that the input is transcribed as text instead of speech. 6.1
Work ow of this system is described as follows (see, Figure 1). 1. Input a text of user's requests. 2. Perform the morphological analysis for the input text. 3. Perform the case analysis starting with surface cases to deep cases. 4. Translate the requests into SPARQL queries. 5. Obtain the reply of SPARQL queries. 6. Generate the answering text from the obtained reply. 7. Output the recommendation.
Japanese is a kind of agglutinative languages and a Japanese sentence is written without a space left among phrases and words. A noun phrase is composed of a noun and a particle, a verb phrase is composed of a stem of verb and a grammatical conjugation. So, morphological analysis is requisite in Japanese text processing in order to separate a sentence into phrases and words. Furthermore, particles attached to nouns decide the grammar case. For example, in response to an user's input \ (I want to see a lion)", the morphological analysis and shift-reduce method changes the Japanese sentence into the form of (( (pos info) 8) (( (pos info) 6) ( (pos info) 5)) (( (pos info) 4) ( (pos info) 0))), here (pos info) stands for a Part-of-Speech information of each, then case analysis produces the result such as Subject:NIL, Verb:( (pos info) 5), Object:( (pos info) 0), toPlace:NIL, fromPlace:NIL, Tool:NIL. Part-of-speech information obtained from morphological analysis is effectively used in various ways. For example, if there is an auxiliary verb ` (want)' next to a form of a behavioral verb such as ` (see)' or ` (eat)', the whole sentence is interpreted as request. Thus, a request of seeing a lion is captured and transformed into a SPARQL query to the endpoint.
From the interpretation of request (see lion), the system searches facilities that can see a lion, using action target ontology and facility ontology. However, we have no common sense as LOD that a lion is in a zoo. When searching fails here, WordNet is used to generalize the target to more abstract ones by searching hypernym relations in WordNet until animal is found.
The SPARQL search picks up a number of facilities that are located near the current location, and the closest one to the current location is chosen outside of SPARQL search. 6.2
Initially, we attempted to make a plan by introducing IS-A logic function into
planning based on classical state space reasoning and backward reasoning [
The following shows an example of execution by this prototype system, see the added comments translated into English for readers.
SYSTEM(4): (eliza) system> ;; I want to enjoy some sport, after that, I want to go to hot spring. ;; the current location is Toyota Higashfuji Institute.
; searching a location for sports ......
; guiding the nearest place 13.37621km ; the distance is 13.37621km
; place: Numazu City Ball Park 35.1125 ; longitude 138.863 ; latitude URL "http://linkedgeodata.org/triplify/node2877270449" (35.1125 . 138.863) ; the current location is (35.1125 . 138.863) ; searching a location for hot spring ......
; guiding the nearest place 10.426165km ; the distance is 10.426165km
; place: Izu-Nagaoka Hot Spring 35.0353 ; longitude 138.929 ; latitude URL "http://ja.dbpedia.org/resource/ "
Searching for a facility in the vicinity of the current location, the Toyota Higashifuji Institute, the system made a recommendation to go to Numazu City Ball Park, then go to Izu-Nagaoka Hot Spring, in response to a request to go to a hot spring after enjoying some sport.
Here the command `eliza' is named for just representing a mimic of Eliza
dialog system [
In this work, we captured knowledge about our world into three layers, i,e., factual knowledge, general knowledge, and empirical knowledge. The factual knowledge includes objective information on individual events and matters. On the other hand, the general knowledge is not information on individual events and things, but rather description of relationships among them in addition to the abstract descriptions of events and things. It is regarded as objectively valid by most people or as social agreement. The empirical knowledge is a speci c knowledge that does not go into general knowledge in society, such as personal knowledge which are agreed only by less people. For example, suppose a very delicious hamburger made by a fast food shop located at a place, the information on this shop's address is factual knowledge, the knowledge of classi cation on fast food shop is general knowledge, and knowledge such as a hamburger made by this hamburger shop is delicious is empirical knowledge.
As shown in Table 1, most of LGD / OSM is factual knowledge and it is categorized to factual dataset, but a part of LGD / OSM is categorized into general knowledge. DBpedia contains both fact data and general knowledge. However, WordNet contains general and empirical common knowledge.
In this preliminary research, the following issues are suggested. 1. It is necessary to understand data characteristics of coverage and granularity of each dataset, but it is generally hard for large datasets. At this time, we rstly made a utilization plan on the whole data set, after we examined the availability of actual data on the premise of these use-case scenarios. 2. Generally, it is tough work to nd out correct relations between datasets.
While simple string matching allows us an automatic matching process, the ontology mapping cannot be avoid human power at the present. While the accuracy of this mapping greatly affects the result, mechanical matching processing is difficult. In addition, we built intermediate ontologies and mapped them to LOD datasets, but building ontology is generally not easy for a novice. 3. Since DBpedia and LGD are datasets made by crowd sourcing, we cannot expect the completeness and validity of them. Missing or biased data is still problematic at reasoning. Actually, we found a closed food shop as results. At this time we attempted to eliminate errors as soon as it was found, but we need to think about some tools for (semi) automated error checking. 4. The inference procedure was designed according to these use-case scenarios.
For other problems, different datasets and different work ows may be used. For example, it depends on features of a target problem about how the balance should be taken between general knowledge and fact data to solve the problem. 9
In this preliminary research, we made a prototype of action planning system for events of everyday life and world, based on open knowledge of LOD as fact data and taxonomy as common knowledge. We utilized a number of large-scale open databases and knowledge-bases. We found that we had already abundant knowledge about the everyday life and world as diverse open knowledge resources. This condition is very different at the era of Good-Old-Fashioned-AI (GOGAI) before the Web age and LOD. However, we also found that we needed the additional general and common knowledge that connects such different open resources in reasoning action plans with SPARQL endpoints. It is obvious that it will be necessary to make open knowledge more available not only in the veri cation and validation for each, but also in the combinations of them for applications.