Jared Daniel Kaplan is a theoretical physicist and artificial intelligence researcher. He is an associate professor in the Johns Hopkins University Department of Physics & Astronomy, and a co-founder and chief science officer of Anthropic. == Education == Kaplan attended the Illinois Mathematics and Science Academy during high school. He received a bachelor's degree in physics and mathematics from Stanford University and a PhD in physics from Harvard University. His doctoral thesis is titled Aspects of holography, advised by Nima Arkani-Hamed. == Academic career and physics research == Kaplan’s research interests include quantum gravity, holography (AdS/CFT), conformal field theory, and related topics in particle physics and cosmology. He worked as a postdoctoral fellow at SLAC and Stanford University and has been a professor at Johns Hopkins University since 2012. == Machine learning research == Kaplan joined OpenAI in 2019 as a researcher, where he co-authored Scaling Laws for Neural Language Models (2020), which reported that empirically, the performance of language models steadily improves with their size and the amount of data and compute used for training. He is also a co-author of Language Models are Few-Shot Learners (2020), which introduced GPT-3. At the company, he was also involved in the development of Codex. == Anthropic == Kaplan co-founded Anthropic and serves as its chief science officer. In October 2024, Anthropic announced that Kaplan would serve as the company's "Responsible Scaling Officer", overseeing its responsible scaling policy (RSP). In this role, Kaplan determines the safety assessments and precautions to adopt before model release. In December 2025, The Guardian published an interview with Kaplan about AI autonomy and recursive self-improvement timelines. == Honors and recognition == Kaplan was a Hertz Fellow (2005). He has also received a Sloan Research Fellowship and an NSF CAREER award (PHY-1454083). == Selected works == Scaling Laws for Neural Language Models (2020). Language Models are Few-Shot Learners (2020). A Natural Language for AdS/CFT Correlators (2011). == Personal life == As of 2026, Forbes estimated Kaplan's net worth at $3.7 billion. He lives in Pacifica, California, and has a son.
Software construction
Software construction is the process of creating working software via coding and integration. The process includes unit and integration testing although does not include higher level testing such as system testing. Construction is an aspect of the software development lifecycle and is integrated in the various software development process models with varying focus on construction as an activity separate from other activities. In the waterfall model, a software development effort consists of sequential phases including requirements analysis, design, and planning which are prerequisites for starting construction. In an iterative model such as scrum, evolutionary prototyping, or extreme programming, construction as an activity that occurs concurrently or overlapping other activities. Construction planning may include defining the order in which components are created and integrated, the software quality management processes, and the allocation of tasks to teams and developers. To facilitate project management, numerous construction aspects can be measured; these include the amount of code developed, modified, reused, and destroyed, code complexity, code inspection statistics, faults-fixed and faults-found rates, and effort expended. These measurements can be useful for aspects such as ensuring quality and improving the process. == Activities == Construction includes many activities. === Coding === The following are a few of the key aspects of the coding activity: Naming Choice of name for each identifier. One study showed that the effort required to debug a program is minimized when variable names are between 10 and 16 characters. Logic Organization into statements and routines Highly cohesive routines proved to be less error prone than routines with lower cohesion. A study of 450 routines found that 50 percent of the highly cohesive routines were fault free compared to only 18 percent of routines with low cohesion. Another study of a different 450 routines found that routines with the highest coupling-to-cohesion ratios had 7 times as many errors as those with the lowest coupling-to-cohesion ratios and were 20 times as costly to fix. Although studies showed inconclusive results regarding the correlation between routine sizes and the rate of errors in them, but one study found that routines with fewer than 143 lines of code were 2.4 times less expensive to fix than larger routines. Another study showed that the code needed to be changed least when routines averaged 100 to 150 lines of code. Another study found that structural complexity and amount of data in a routine were correlated with errors regardless of its size. Interfaces between routines are some of the most error-prone areas of a program. One study showed that 39 percent of all errors were errors in communication between routines. Unused parameters are correlated with an increased error rate. In one study, only 17 to 29 percent of routines with more than one unreferenced variable had no errors, compared to 46 percent in routines with no unused variables. The number of parameters of a routine should be 7 at maximum as research has found that people generally cannot keep track of more than about seven chunks of information at once. One experiment showed that designs which access arrays sequentially, rather than randomly, result in fewer variables and fewer variable references. One experiment found that loops-with-exit are more comprehensible than other kinds of loops. Regarding the level of nesting in loops and conditionals, studies have shown that programmers have difficulty comprehending more than three levels of nesting. Control flow complexity has been shown to correlate with low reliability and frequent errors. Modularity Structuring and refactoring the code into classes, packages and other structures. When considering containment, the maximum number of data members in a class shouldn't exceed 7±2. Research has shown that this number is the number of discrete items a person can remember while performing other tasks. When considering inheritance, the number of levels in the inheritance tree should be limited. Deep inheritance trees have been found to be significantly associated with increased fault rates. When considering the number of routines in a class, it should be kept as small as possible. A study on C++ programs has found an association between the number of routines and the number of faults. A study by NASA showed that the putting the code into well-factored classes can double the code reusability compared to the code developed using functional design. Error handling Encoding logic to handle both planned and unplanned errors and exceptions. Resource management Managing computational resource use via exclusion mechanisms and discipline in accessing serially reusable resources, including threads or database locks. Security Prevention of code-level security breaches such as buffer overrun and array index overflow. Optimization Optimization while avoiding premature optimization. Documentation Both embedded in the code as comments and as external documents. === Integration === Integration is about combining separately constructed parts. Concerns include planning the sequence in which components will be integrated, creating scaffolding to support interim versions of the software, determining the degree of testing and quality work performed on components before they are integrated, and determining points in the project at which interim versions are tested. === Testing === Testing can reduce the time between when faulty logic is inserted in the code and when it is detected. In some cases, testing is performed after code has been written, but in test-first programming, test cases are created before code is written. Construction includes at least two forms of testing, often performed by the developer who wrote the code: unit testing and integration testing. === Reuse === Software reuse entails more than creating and using libraries. It requires formalizing the practice of reuse by integrating reuse processes and activities into the software life cycle. The tasks related to reuse in software construction during coding and testing may include: selection of the reusable code, evaluation of code or test re-usability, reporting reuse metrics. === Quality assurance === Techniques for ensuring quality as software is constructed include: Testing One study found that the average defect detection rates of Unit testing and integration testing are 30% and 35% respectively. Software inspection With respect to software inspection, one study found that the average defect detection rate of formal code inspections is 60%. Regarding the cost of finding defects, a study found that code reading detected 80% more faults per hour than testing. Another study shown that it costs six times more to detect design defects by using testing than by using inspections. A study by IBM showed that only 3.5 hours were needed to find a defect through code inspections versus 15–25 hours through testing. Microsoft has found that it takes 3 hours to find and fix a defect by using code inspections and 12 hours to find and fix a defect by using testing. In a 700 thousand lines program, it was reported that code reviews were several times as cost-effective as testing. Studies found that inspections result in 20% - 30% fewer defects per 1000 lines of code than less formal review practices and that they increase productivity by about 20%. Formal inspections will usually take 10% - 15% of the project budget and will reduce overall project cost. Researchers found that having more than 2 - 3 reviewers on a formal inspection doesn't increase the number of defects found, although the results seem to vary depending on the kind of material being inspected. Technical review With respect to technical review, one study found that the average defect detection rates of informal code reviews and desk checking are 25% and 40% respectively. Walkthroughs were found to have a defect detection rate of 20% - 40%, but were found also to be expensive especially when project pressures increase. Code reading was found by NASA to detect 3.3 defects per hour of effort versus 1.8 defects per hour for testing. It also finds 20% - 60% more errors over the life of the project than different kinds of testing. A study of 13 reviews about review meetings, found that 90% of the defects were found in preparation for the review meeting while only around 10% were found during the meeting. Static analysis With respect to Static analysis (IEEE1028), studies have shown that a combination of these techniques needs to be used to achieve a high defect detection rate. Other studies showed that different people tend to find different defects. One study found that the extreme programming practices of pair programming, desk checking, unit testing, integration testing, and regression testing can achieve a 90% defect detection rate. An experiment involving exper
H2O (software)
H2O is an open-source, in-memory, distributed machine learning and predictive analytics platform developed by the company H2O.ai (previously 0xdata). The software uses a distributed architecture for parallel processing on standard hardware. It supports algorithms for large-scale data analysis and model deployment. H2O is primarily used by data scientists and developers for statistical modeling and data-driven decision-making. The platform is designed to handle in-memory computations across a distributed computing environment. It offers implementations for numerous statistical and machine learning algorithms, which are accessible through various programming interfaces. The software is released under the Apache License 2.0. == Functionality and features == H2O provides a suite of supervised and unsupervised machine learning algorithms. Its core functions include: Supervised learning: algorithms in the field of statistics, data mining and machine learning such as generalized linear models, random forests, gradient boosting and deep learning are implemented for classification and regression tasks. Unsupervised learning: including K-Means clustering and principal component analysis. Automated machine learning: a features designed to automate the processes of model selection, tuning, and ensemble creation. The software can ingest data from various sources, including the Hadoop Distributed File System, Amazon S3, SQL databases, as well as local file systems. It operates natively on Apache Spark clusters through Sparkling Water. Proponents claim that improved performance is achieved compared to other analysis tools. The software is distributed free of charge, under a business model based on the development of individual applications and support. == Architecture == H2O is primarily written in Java. It uses a distributed architecture that allows the platform to cluster nodes for parallel processing and in-memory storage of data and models. Users interact with the H2O platform through several primary interfaces: Programming language interfaces: APIs are provided for the R and Python programming languages, and various Apache offerings (Apache Hadoop and Spark, as well as Maven). H2O Flow: a graphical web-based interactive computational environment that functions as a notebook interface for data exploration, model building, and scripting. REST-API: allows for integration with other applications and frameworks such as Microsoft Excel or RStudio. With the H2O Machine Learning Integration Nodes, KNIME offers algorithmic workflows. While the algorithm executes, approximate results are displayed, so that users can track the progress and intervene if needed. == History, influences, and extensions == The software project was initiated by the company 0xdata, which later changed its name to H2O.ai. The three Stanford professors Stephen P. Boyd, Robert Tibshirani and Trevor Hastie form a panel that advises H2O on scientific issues. Since its inception, H2O provides open-source machine learning libraries for enterprise use. The core H2O platform is often complemented by offerings from H2O.ai, such as H2O Driverless AI. == Reception == H2O is referenced in peer-reviewed literature regarding automated machine learning (AutoML). The platform has been categorized as a "Leader" and a "Strong Performer" in industry reports by Forrester Research. H2O (the open-source platform) and the associated commercial platform Driverless AI have been recurring winners of InfoWorld's most prestigious awards, including both the Best of Open Source Software ("Bossies") and the Technology of the Year awards.
Gibberlink
GibberLink is an acoustic data transmission project, with an open-source client available on GitHub, in which two conversational AI agents switch from speaking to one another in a Human-listenable language (such as English) to their own unique language that consists of a sound-level protocol after confirming they are both AI agents. The project was created by Anton Pidkuiko and Boris Starkov. == Reception == The project won the global top prize at the ElevenLabs Worldwide Hackathon. It has also been cited as raising questions around AI ethics and oversight. On February 23, 2025, a YouTube video of two independent conversational ElevenLabs AI agents being prompted to chat about booking a hotel (one as a caller, one as a receptionist) received coverage for going viral. In this video, both agents are prompted to switch to ggwave data-over-sound protocol when they identify the other side as AI, and keep speaking in English otherwise.
Representation collapse
Representation collapse is a phenomenon in machine learning and representation learning where a model maps different inputs to the same or very similar embeddings, which means it loses important information about how the data is spread out. It is frequently encountered in self-supervised learning, especially within contrastive and non-contrastive frameworks, when training objectives or model architectures do not maintain variance across representations. Collapse results in degenerate solutions characterized by uninformative learned features, significantly impairing downstream task performance. Various techniques have been proposed to mitigate representation collapse, including the use of negative samples, architectural asymmetry, stop-gradient operations, variance regularization, and redundancy reduction objectives, as seen in methods such as SimCLR, BYOL, and VICReg. Comprehending and averting representation collapse is regarded as a fundamental challenge in the advancement of stable and efficient self-supervised learning systems.
Contrastive Language-Image Pre-training
Contrastive Language-Image Pre-training (CLIP) is a technique for training a pair of neural network models, one for image understanding and one for text understanding, using a contrastive objective. This method has enabled broad applications across multiple domains, including cross-modal retrieval, text-to-image generation, and aesthetic ranking. == Algorithm == The CLIP method trains a pair of models contrastively. One model takes in a piece of text as input and outputs a single vector representing its semantic content. The other model takes in an image and similarly outputs a single vector representing its visual content. The models are trained so that the vectors corresponding to semantically similar text-image pairs are close together in the shared vector space, while those corresponding to dissimilar pairs are far apart. To train a pair of CLIP models, one would start by preparing a large dataset of image-caption pairs. During training, the models are presented with batches of N {\displaystyle N} image-caption pairs. Let the outputs from the text and image models be respectively v 1 , . . . , v N , w 1 , . . . , w N {\displaystyle v_{1},...,v_{N},w_{1},...,w_{N}} . Two vectors are considered "similar" if their dot product is large. The loss incurred on this batch is the multi-class N-pair loss, which is a symmetric cross-entropy loss over similarity scores: − 1 N ∑ i ln e v i ⋅ w i / T ∑ j e v i ⋅ w j / T − 1 N ∑ j ln e v j ⋅ w j / T ∑ i e v i ⋅ w j / T {\displaystyle -{\frac {1}{N}}\sum _{i}\ln {\frac {e^{v_{i}\cdot w_{i}/T}}{\sum _{j}e^{v_{i}\cdot w_{j}/T}}}-{\frac {1}{N}}\sum _{j}\ln {\frac {e^{v_{j}\cdot w_{j}/T}}{\sum _{i}e^{v_{i}\cdot w_{j}/T}}}} In essence, this loss function encourages the dot product between matching image and text vectors ( v i ⋅ w i {\displaystyle v_{i}\cdot w_{i}} ) to be high, while discouraging high dot products between non-matching pairs. The parameter T > 0 {\displaystyle T>0} is the temperature, which is parameterized in the original CLIP model as T = e − τ {\displaystyle T=e^{-\tau }} where τ ∈ R {\displaystyle \tau \in \mathbb {R} } is a learned parameter. Other loss functions are possible. For example, Sigmoid CLIP (SigLIP) proposes the following loss function: L = 1 N ∑ i , j ∈ 1 : N f ( ( 2 δ i , j − 1 ) ( e τ w i ⋅ v j + b ) ) {\displaystyle L={\frac {1}{N}}\sum _{i,j\in 1:N}f((2\delta _{i,j}-1)(e^{\tau }w_{i}\cdot v_{j}+b))} where f ( x ) = ln ( 1 + e − x ) {\displaystyle f(x)=\ln(1+e^{-x})} is the negative log sigmoid loss, and the Dirac delta symbol δ i , j {\displaystyle \delta _{i,j}} is 1 if i = j {\displaystyle i=j} else 0. == CLIP models == While the original model was developed by OpenAI, subsequent models have been trained by other organizations as well. === Image model === The image encoding models used in CLIP are typically vision transformers (ViT). The naming convention for these models often reflects the specific ViT architecture used. For instance, "ViT-L/14" means a "vision transformer large" (compared to other models in the same series) with a patch size of 14, meaning that the image is divided into 14-by-14 pixel patches before being processed by the transformer. The size indicator ranges from B, L, H, G (base, large, huge, giant), in that order. Other than ViT, the image model is typically a convolutional neural network, such as ResNet (in the original series by OpenAI), or ConvNeXt (in the OpenCLIP model series by LAION). Since the output vectors of the image model and the text model must have exactly the same length, both the image model and the text model have fixed-length vector outputs, which in the original report is called "embedding dimension". For example, in the original OpenAI model, the ResNet models have embedding dimensions ranging from 512 to 1024, and for the ViTs, from 512 to 768. Its implementation of ViT was the same as the original one, with one modification: after position embeddings are added to the initial patch embeddings, there is a LayerNorm. Its implementation of ResNet was the same as the original one, with 3 modifications: In the start of the CNN (the "stem"), they used three stacked 3x3 convolutions instead of a single 7x7 convolution, as suggested by. There is an average pooling of stride 2 at the start of each downsampling convolutional layer (they called it rect-2 blur pooling according to the terminology of ). This has the effect of blurring images before downsampling, for antialiasing. The final convolutional layer is followed by a multiheaded attention pooling. ALIGN a model with similar capabilities, trained by researchers from Google used EfficientNet, a kind of convolutional neural network. === Text model === The text encoding models used in CLIP are typically Transformers. In the original OpenAI report, they reported using a Transformer (63M-parameter, 12-layer, 512-wide, 8 attention heads) with lower-cased byte pair encoding (BPE) with 49152 vocabulary size. Context length was capped at 76 for efficiency. Like GPT, it was decoder-only, with only causally-masked self-attention. Its architecture is the same as GPT-2. Like BERT, the text sequence is bracketed by two special tokens [SOS] and [EOS] ("start of sequence" and "end of sequence"). Take the activations of the highest layer of the transformer on the [EOS], apply LayerNorm, then a final linear map. This is the text encoding of the input sequence. The final linear map has output dimension equal to the embedding dimension of whatever image encoder it is paired with. These models all had context length 77 and vocabulary size 49408. ALIGN used BERT of various sizes. == Dataset == === WebImageText === The CLIP models released by OpenAI were trained on a dataset called "WebImageText" (WIT) containing 400 million pairs of images and their corresponding captions scraped from the internet. The total number of words in this dataset is similar in scale to the WebText dataset used for training GPT-2, which contains about 40 gigabytes of text data. The dataset contains 500,000 text-queries, with up to 20,000 (image, text) pairs per query. The text-queries were generated by starting with all words occurring at least 100 times in English Wikipedia, then extended by bigrams with high mutual information, names of all Wikipedia articles above a certain search volume, and WordNet synsets. The dataset is private and has not been released to the public, and there is no further information on it. ==== Data preprocessing ==== For the CLIP image models, the input images are preprocessed by first dividing each of the R, G, B values of an image by the maximum possible value, so that these values fall between 0 and 1, then subtracting by [0.48145466, 0.4578275, 0.40821073], and dividing by [0.26862954, 0.26130258, 0.27577711]. The rationale was that these are the mean and standard deviations of the images in the WebImageText dataset, so this preprocessing step roughly whitens the image tensor. These numbers slightly differ from the standard preprocessing for ImageNet, which uses [0.485, 0.456, 0.406] and [0.229, 0.224, 0.225]. If the input image does not have the same resolution as the native resolution (224×224 for all except ViT-L/14@336px, which has 336×336 resolution), then the input image is first scaled by bicubic interpolation, so that its shorter side is the same as the native resolution, then the central square of the image is cropped out. === Others === ALIGN used over one billion image-text pairs, obtained by extracting images and their alt-tags from online crawling. The method was described as similar to how the Conceptual Captions dataset was constructed, but instead of complex filtering, they only applied a frequency-based filtering. Later models trained by other organizations had published datasets. For example, LAION trained OpenCLIP with published datasets LAION-400M, LAION-2B, and DataComp-1B. == Training == In the original OpenAI CLIP report, they reported training 5 ResNet and 3 ViT (ViT-B/32, ViT-B/16, ViT-L/14). Each was trained for 32 epochs. The largest ResNet model took 18 days to train on 592 V100 GPUs. The largest ViT model took 12 days on 256 V100 GPUs. All ViT models were trained on 224×224 image resolution. The ViT-L/14 was then boosted to 336×336 resolution by FixRes, resulting in a model. They found this was the best-performing model. In the OpenCLIP series, the ViT-L/14 model was trained on 384 A100 GPUs on the LAION-2B dataset, for 160 epochs for a total of 32B samples seen. == Applications == === Cross-modal retrieval === CLIP's cross-modal retrieval enables the alignment of visual and textual data in a shared latent space, allowing users to retrieve images based on text descriptions and vice versa, without the need for explicit image annotations. In text-to-image retrieval, users input descriptive text, and CLIP retrieves images with matching embeddings. In image-to-text retrieval, images are used to find related text content. CLIP’s ability to connect vis
Coupled pattern learner
Coupled Pattern Learner (CPL) is a machine learning algorithm which couples the semi-supervised learning of categories and relations to forestall the problem of semantic drift associated with boot-strap learning methods. == Coupled Pattern Learner == Semi-supervised learning approaches using a small number of labeled examples with many unlabeled examples are usually unreliable as they produce an internally consistent, but incorrect set of extractions. CPL solves this problem by simultaneously learning classifiers for many different categories and relations in the presence of an ontology defining constraints that couple the training of these classifiers. It was introduced by Andrew Carlson, Justin Betteridge, Estevam R. Hruschka Jr. and Tom M. Mitchell in 2009. == CPL overview == CPL is an approach to semi-supervised learning that yields more accurate results by coupling the training of many information extractors. Basic idea behind CPL is that semi-supervised training of a single type of extractor such as ‘coach’ is much more difficult than simultaneously training many extractors that cover a variety of inter-related entity and relation types. Using prior knowledge about the relationships between these different entities and relations CPL makes unlabeled data as a useful constraint during training. For e.g., ‘coach(x)’ implies ‘person(x)’ and ‘not sport(x)’. == CPL description == === Coupling of predicates === CPL primarily relies on the notion of coupling the learning of multiple functions so as to constrain the semi-supervised learning problem. CPL constrains the learned function in two ways. Sharing among same-arity predicates according to logical relations Relation argument type-checking === Sharing among same-arity predicates === Each predicate P in the ontology has a list of other same-arity predicates with which P is mutually exclusive. If A is mutually exclusive with predicate B, A’s positive instances and patterns become negative instances and negative patterns for B. For example, if ‘city’, having an instance ‘Boston’ and a pattern ‘mayor of arg1’, is mutually exclusive with ‘scientist’, then ‘Boston’ and ‘mayor of arg1’ will become a negative instance and a negative pattern respectively for ‘scientist.’ Further, Some categories are declared to be a subset of another category. For e.g., ‘athlete’ is a subset of ‘person’. === Relation argument type-checking === This is a type checking information used to couple the learning of relations and categories. For example, the arguments of the ‘ceoOf’ relation are declared to be of the categories ‘person’ and ‘company’. CPL does not promote a pair of noun phrases as an instance of a relation unless the two noun phrases are classified as belonging to the correct argument types. === Algorithm description === Following is a quick summary of the CPL algorithm. Input: An ontology O, and a text corpus C Output: Trusted instances/patterns for each predicate for i=1,2,...,∞ do foreach predicate p in O do EXTRACT candidate instances/contextual patterns using recently promoted patterns/instances; FILTER candidates that violate coupling; RANK candidate instances/patterns; PROMOTE top candidates; end end ==== Inputs ==== A large corpus of Part-Of-Speech tagged sentences and an initial ontology with predefined categories, relations, mutually exclusive relationships between same-arity predicates, subset relationships between some categories, seed instances for all predicates, and seed patterns for the categories. ==== Candidate extraction ==== CPL finds new candidate instances by using newly promoted patterns to extract the noun phrases that co-occur with those patterns in the text corpus. CPL extracts, Category Instances Category Patterns Relation Instances Relation Patterns ==== Candidate filtering ==== Candidate instances and patterns are filtered to maintain high precision, and to avoid extremely specific patterns. An instance is only considered for assessment if it co-occurs with at least two promoted patterns in the text corpus, and if its co-occurrence count with all promoted patterns is at least three times greater than its co-occurrence count with negative patterns. ==== Candidate ranking ==== CPL ranks candidate instances using the number of promoted patterns that they co-occur with so that candidates that occur with more patterns are ranked higher. Patterns are ranked using an estimate of the precision of each pattern. ==== Candidate promotion ==== CPL ranks the candidates according to their assessment scores and promotes at most 100 instances and 5 patterns for each predicate. Instances and patterns are only promoted if they co-occur with at least two promoted patterns or instances, respectively. == Meta-Bootstrap Learner == Meta-Bootstrap Learner (MBL) was also proposed by the authors of CPL. Meta-Bootstrap learner couples the training of multiple extraction techniques with a multi-view constraint, which requires the extractors to agree. It makes addition of coupling constraints on top of existing extraction algorithms, while treating them as black boxes, feasible. MBL assumes that the errors made by different extraction techniques are independent. Following is a quick summary of MBL. Input: An ontology O, a set of extractors ε Output: Trusted instances for each predicate for i=1,2,...,∞ do foreach predicate p in O do foreach extractor e in ε do Extract new candidates for p using e with recently promoted instances; end FILTER candidates that violate mutual-exclusion or type-checking constraints; PROMOTE candidates that were extracted by all extractors; end end Subordinate algorithms used with MBL do not promote any instance on their own, they report the evidence about each candidate to MBL and MBL is responsible for promoting instances. == Applications == In their paper authors have presented results showing the potential of CPL to contribute new facts to existing repository of semantic knowledge, Freebase