• No se han encontrado resultados

Valoración teórica de la efectividad del desarrollo de la apreciación de la arquitectura local en los escolares de sexto grado de la Educación Primaria.

CAPITULO III. ESTRATEGIA DIDÁCTICA PARA EL DESARROLLO DEL PROCESO DE APRECIACIÓN DE LA ARQUITECTURA LOCAL EN LOS ESCOLARES DE SEXTO GRADO DE LA EDUCACIÓN PRIMARIA

INSTRUMENTACIÓN Objetivo

3.3. Valoración teórica de la efectividad del desarrollo de la apreciación de la arquitectura local en los escolares de sexto grado de la Educación Primaria.

Next, we survey declarative languages that are available for querying different aspects of the Twittersphere paying attention to their underlying data models and query dimensions. The goal of proposing declarative languages and systems for querying tweets is to put forward a set of primitives or an interface for analysts to conveniently query specific interactions on Twitter,

exploring the user, time, space and topical dimensions. High level languages for querying

tweets extend the capabilities of existing languages such as SQL and SPARQL. Queries are either executed on the Twitter stream in real-time or on a stored off-line collection.

Languages for Querying Tweets 39

3.4.1 Generic Languages

TweeQL [128] provides a streaming SQL-like interface to the Twitter API and provides a set

of user-defined functions (UDFs) to manipulate data. The objective is to introduce a query language to extract structure and useful information embedded in unstructured Twitter data. The language exploits both relational and streaming semantics. UDFs allow for operations such as location identification, string processing, sentiment prediction, named entity extraction and event detection. In the spirit of streaming semantics, it provides SQL constructs to perform aggregations over the incoming stream on a user-specified time window. The result of a given query can be stored in a relational fashion for subsequent querying.

Models for representing any social network in RDF have been proposed by Martin and

Gutierrez [129] allowing queries in SPARQL. Their work explored the feasibility of adoption of

this model by demonstrating their idea with an illustrative prototype but did not focus on a

single social network such as Twitter in particular. TwarQL [135] extracts content from tweets

and encodes it in RDF format using shared and well-known vocabularies (FOAF, MOAT, SIOC) enabling querying in SPARQL. The extraction facility processes plain tweets and expands its description by adding sentiment annotations, DBPedia entities, hashtag definitions and URLs. The annotation of tweets using different vocabularies enables querying and analysis in different dimensions such as location, users, sentiment and related named entities.

Temporal and topical features are of paramount importance in an evolving microblogging stream like Twitter. In the languages above, time and topic of a tweet (topic can be represented simply by a hashtag) are considered meta-data of the tweet and are not treated any differently from other metadata reported. Topics are regarded as part of the tweet content or what drives the data filtering task from the Twitter API. There have been efforts to exploit features that

go well beyond a simple filter based on time and topic. Plachouras and Stavrakas [154] stressed

the need for temporal modeling of terms in Twitter to effectively capture changing trends. A term refers to any word or short phrase of interest in a tweet, including hashtags or output of an entity recognition process. Their proposed query operators can express complex queries for associations between terms over varying time granularities, to discover the context of collected data. Operators also allow retrieving a subset of tweets satisfying these complex conditions on term associations. This enables the end-user to select a good set of terms (hashtags) that drive the data collection, and this has a direct impact on the quality of the results generated by the analysis.

Languages for Querying Tweets 40

Spatial features are another property of tweets often overlooked in complex analyses. Previ- ously discussed studies use the location attribute as a mechanism to filter tweets. To complete our discussion, we briefly outline two studies that used geo-spatial properties to perform com-

plex analysis using the location attribute. Doytsher et al.[53] introduced a model and query

language suited for integrated data connecting a social network of users with a spatial network to identify places visited frequently. Edges named life-patterns are used to associate the social and spatial networks. Different time granularities can be expressed for each visited location represented by the life-pattern edge. Even though the implementation employs a partially synthetic dataset, it will be interesting to investigate how the socio-spatial networks and the life-pattern edges that are used to associate the spatial and social networks can be represented

in a real social network dataset with location information, such as Twitter. GeoScope [26] finds

information trends by detecting significant correlations among trending location-topic pairs in a sliding window. This gives rise to the importance of capturing the notion of spatial information trends in social networks in analysis tasks. Real-time detection of crisis events from a location in space, exhibits the possible value of Geoscope. In one of the experiments, Twitter is used as a case study to demonstrate its usefulness: a hashtag is chosen to represent the topic and city from which the tweet originates is chosen to capture the location.

3.4.2 Query Languages for Social Networks

To the best of our knowledge, there is no existing work focusing on high level languages operating on the Twitter’s social graph. However it is important to note proposals for declarative query

languages tailored for querying social networks in general [5,179,54,129,130,168]. One of the

queries supported are path queries satisfying a set of conditions on the path, and the languages in general take advantage of inherent properties of social networks. Semantics of the languages

are based on Datalog [130], SQL [54,168] or SPARQL [129]. Implementations are conducted

on bibliographical networks [54], Facebook and social content sites like Yahoo! Travel [5] and

are not tested on Twitter networks taking Twitter specific affordances into consideration.

3.4.3 Information Retrieval - Tweet Search

Another class of systems presents textual queries to efficiently search over a corpus of tweets. The challenges in this area are similar to that of information retrieval but also have to deal with peculiarities of tweets. The short length of tweets in particular creates added complexity to text-based search tasks as it is difficult to identify relevant tweets matching a user query

Languages for Querying Tweets 41

[16,66]. Expanding tweet content is suggested as a way to enhance meaning. The goal of such

systems is to express a user’s information need in the form of a text query, much as in search engines, and return a tweet list in real-time with effective strategies for ranking and relevance

measurements [55,63,209]. Indexing mechanisms were discussed in [36] as they directly impact

efficient retrieval of tweets. The TREC microblogging track6is dedicated to calling participants

to conduct real-time ad-hoc search tasks over a given tweet collection. Publications of TREC

[149] document the findings of all systems in the task of ranking the most relevant tweets

matching a pre-defined set of user-queries.

3.4.4 Discussion: Data Model and Storage for the Languages

Relational, RDF and Graphs are the most common choices of data representation. There is a close affiliation in these data models observing that, for instance, a graph can correspond to a

set of RDF triples or vice versa. In fact, some studies like Plachouras and Stavrakas [154] have

put forward their data model as a labeled multi digraph and have chosen a relational database for their implementation. None of these query systems models Twitter social network with

following or retweet relationships among users. Doytsher et al.[53] implemented their algebraic

query operators with the use of both graph and a relational database as the underlying data storage. They experimentally compared relational and graph database systems to demonstrate the feasibility of the model. Languages that operate on the twitter stream such as TweeQL

and TwarQL generate the output in real-time; TweeQL [128] allows the resulting tweets to be

collected in batches then stores them in a relational database, while TwarQL [135] at the end

of the information extraction phase, encodes annotated tweets in RDF.

Table 3.2: Overview of approaches in systems for querying tweets.

Data Model Explored dimensions

Relational RDF Graph Text Time Space Social Network Real-Time

TweeQL [128] X X X X Yes

TwarQL [135] X X X X Yes

Plachouras et al.[155] X X XX No

Doytsher et al.[53]∗ X X X XX X No

GeoScope et al.[26]∗ X X X X Yes

Languages on social networks∗ X X X X XX No

Tweet search systems X XX X X Yes

Table 3.2illustrates an overview of related approaches in systems for querying tweets. Data models and dimensions investigated in each system are depicted. Systems that have made

Requirements of an Integrated Solution 42

provision for the real-time streaming nature of the tweets are indicated in the Real-time column.

Multiple ticks (X) correspond to a dimension explored in detail. Note that the systems marked

with an asterisk (*) are not implemented specifically targeting tweets, although their application is meaningful and can be extended to the Twittersphere. We observe there is a potential for developing languages for querying tweets that include querying by dimensions that are not captured by existing systems, especially the social graph.